Genetic variation in cultivated coffee (Coffea arabica L.) accessions in northern New South Wales, Australia

Transcription

1 Southern Cross University Theses 2005 Genetic variation in cultivated coffee (Coffea arabica L.) accessions in northern New South Wales, Australia Thi Minh Hue Tran Southern Cross University Publication details Hue, TMH 2005, 'Genetic variation in cultivated coffee (Coffea arabica L.) accessions in northern New South Wales, Australia', Masters thesis, Southern Cross University, Lismore, NSW. Copyright TMH Tran 2005 is an electronic repository administered by Southern Cross University Library. Its goal is to capture and preserve the intellectual output of Southern Cross University authors and researchers, and to increase visibility and impact through open access to researchers around the world. For further information please contact

2 GENETIC VARIATION IN CULTIVATED COFFEE (Coffea arabica L.) ACCESSIONS IN NORTHERN NEW SOUTH WALES, AUSTRALIA By Tran Thi Minh Hue BSc Centre for Plant Conservation Genetics Thesis submitted for the fulfillment of the requirements for the Degree of Master of Science at Southern Cross University December 2005

3 Declaration I, Tran Thi Minh Hue, declare that the work presented in the following thesis, to the best of my knowledge and belief is original, except as acknowledged in the text. The material has not been submitted, or being considered for submission, either in whole or in part, for another degree at this or any other university.. Tran Thi Minh Hue i

4 Acknowledgements First, I would like to acknowledge the following people involved in my completion of the degree of Master of Science. My Principal supervisor, Associate Professor Slade Lee, for the supervision and guidance from the very first days till the end of the study; and for sustained enthusiasm and patience throughout all phases of the project. I extend my very warmest and most sincere thanks to him for bringing me not only the knowledge but also the confidence and encouragement. Dr. Martin Elphinstone, Dr. Merv Shepherd, Dr. Nicole Rice and Dr. Dan Waters for checking my writing and giving valuable suggestions. Natalie Baker PhD student for assisting me in editing my writing, analysing data and giving critical comments. Shane McIntosh PhD student, Linh Nguyen PhD student, Dr. Agnelo Furtado, Stirling Bowen, Adam Benson, Dr. Liz Izquierdo for assisting my technical work. And all staff and students of the Centre for Plant Conservation Genetics for encouragement and support. AusAID for financial support, a vital constituent for my study. The Vietnamese Ministry of Education and Training and The Vietnamese Ministry of Agriculture and Rural Development for permission. Dr. Hoang Thanh Tiem for permission and support in my study, and for authorisation on collecting samples. Dr. Trinh Duc Minh for his advice, encouragement and moral support in my study. MSc Vuong Phan for permission and support during my study. My colleagues in Vietnam (MSc Tran Anh Hung, Dinh Thi Tieu Oanh, Nguyen Thi Thanh Mai and Tran Thi Ngoc Lan) and local coffee growers in NNSW (Ian Cannon and Dale Potts, John Zentveld, Steve Myers and Wendy King) for providing samples. My colleagues of the Western Highlands Agro-Forestry Science and Technical Institute (WASI) for encouragement and support. ii

5 My friend Nguyen Vu Ky for helping me with IT issues. Secondly, I am grateful to My parents-in-law for their sharing in looking after my son with my husband and their understanding and endless encouragement, My parents for considerate understanding and support, The rest of my family, my siblings and in-laws, especially my sister Tran Thi Thanh Huyen, my brother in-law Tran Phuc Hau and my sister-in-law Tran Thi Kim Loan for moral support. Finally but most importantly and decisively, my beloved husband Tran Anh Hung for his endless encouragement, patience and sacrifice when taking care of our son Tran Anh Minh while I am studying in Southern Cross University; my cherished son for his tolerance of being far away from his mum. I am indebted to all of you. iii

6 Abstract Genetic consistency within varieties is essential to quality assurance for any agricultural product. While the Australian coffee industry targets high quality coffee, there is observed morphological variation within coffee varieties in New South Wales plantations. This variability may result from environmental, genetic and/or management factors. Genetic factors can be tested by molecular markers which can also shed light on the questions concerning crop quality management. A review of the literature showed low genetic variation in C. arabica. Hence four different molecular marker systems were used in this study to detect possible genetic variation within and between varieties of local coffee grown in Northern New South Wales (NNSW), Australia. Genetic variation in eighty-four seed propagated coffee (C. arabica) accessions, mainly from two commercial varieties (K7 and CRB) in NNSW, were tested using various PCR-based marker systems (RAPDs, ISSRs, SSRs and AFLPs). Eleven accessions from Central Highland, Vietnam, were used as reference material. While RAPD and ISSR did not distinguish intra-varietal molecular variation, SSR and AFLP data revealed the degree of genetic variability and the relationship among individuals within and between coffee varieties. Despite observed morphological variation within supposedly single variety plantations in NNSW, the genetic variation, measured by genetic distance, revealed in this study was very low (K7: 0.193; CRB: 0.205). There exists genetic variation between different farms sharing the same cultivar (K7) which suggests differences in the management of plantation establishment and sourcing of trees. The genetic variability is not aligned with off-type individuals observed in K7, but is with off-type CRB plants which is probably due to inter-varietal hybrids from unintentional outcrossing. The mean level of genetic identity between cultivars derived from the two distinct types of C. arabica is moderate (0.641). Although genetic variation within and among arabica cultivars is low, sufficient DNA polymorphism was found among some C. arabica accessions to allow differentiation. The results in this study suggested that even the elite cultivars, which have been exposed to intensive selection, still show a certain degree of genetic variation amongst individuals within each cultivar even though C. arabica is a predominantly selfing species and has a narrow genetic foundation. The congruence between AFLP and SSR data sets suggests that either method individually, or a combination, is applicable to genetic studies of coffee. SSR alone clearly distinguished and revealed inter-varietal heterogeneity but were more powerful when combined with AFLP.

7 TABLE OF CONTENTS Abstract..0 CHAPTER I INTRODUCTION & LITERATURE REVIEW...1 INTRODUCTION Classification of coffee - taxonomy, origin and early history Importance of coffee to agriculture Coffee in Northern New South Wales (NNSW) Australia Biological questions and hypothesis Morphological diversity vs single variety Objective of the present study.6 LITERATURE REVIEW Coffee breeding Coffee propagation Out-crossing in C. arabica Mutants in C. arabica Biotechnology for coffee improvement Genetic markers for coffee Morphological markers Biochemical markers..12 Isozymes DNA-based molecular markers Hybridisation-based (non-pcr) markers 13 RFLP markers in general...13 RFLP markers in coffee PCR-based markers. 14 RAPD markers RAPD markers in coffee..15 SSR markers SSR markers in coffee.. 17 ISSR markers...19 ISSR markers in coffee. 20 AFLP markers. 21 AFLP markers in coffee...22

8 CHAPTER II - METHODS Sample collection Sampling in New South Wales, Australia Sampling in Daklak, Vietnam DNA extraction Sample storage Different methods of DNA extraction 25 DNA extraction protocol used in this study PCR optimisation RAPD markers Primers and enzymes MgCl 2 concentration Annealing temperatures (Ta) DNA template preparation dntp concentration, DNA polymerase and buffer type PCR conditions ISSR markers Primers used in the study PCR optimisation PCR SSR markers Primer information PCR AFLP markers AFLP adapters and primers DNA digestion and ligation AFLP pre-selective and selective PCRs Detection method RAPD and ISSR markers SSR markers AFLP markers Data scoring and analysis SSR markers AFLP markers 44

9 Combined SSR and AFLP data..44 CHAPTER III RESULTS RAPD results ISSR results SSR markers AFLP markers Combined SSR and AFLP data Correlation between AFLPs and SSRs Genetic variation between varieties Genetic variation within varieties Genetic relationship among varieties...72 CHAPTER IV - DISCUSSION AND CONCLUSION Discussion Marker discrimination power SSR markers AFLP markers Genetic variation among varieties Genetic variation within varieties Conclusions...88 REFERENCES. 90 APPENDIX 1 - Accessions of C. arabica in the study

10 FIGURES & TABLES Fig. 1.1 Schematic representation of the main steps in the history of coffee cultivation (Anthony et al 2002a). Fig. 1.2 Off-type K7 on plantation, Byron Bay, Australia. 5 Fig. 1.3 True-to-type (left) and off-type (right) leaves of K7. 5 Fig. 1.4 Typical leaf colour of CRB and K7. 11 Fig. 1.5 Branching characters of CRB and K7. 12 Fig. 2.1 DNA amplification by PCR indicates the effect of different methods of DNA extraction. Fig. 2.2 Restriction efficiency of five enzymes conducted on four repeated samples. 30 Fig. 2.3 PCR optimization to examine the annealing temperature using ISSR R02 primer. Fig. 2.4 SSR alleles on one locus amplified by CM16 primer pair on the ABI 3730 capillary electrophoresis system. Fig. 2.5 AFLP fragments amplified by the combination of EcoRI-AAC-NED and MseI-CAA on the ABI 3730 capillary electrophoresis system. Fig. 3.1 The banding pattern of primer OPN20 using genomic template DNA digested with different restriction enzymes (each enzyme digested four different DNA samples). Fig. 3.2 The banding pattern of RAPD primer OPN20 using HindIII digested genomic template DNA showing polymorphic locus at 1150 bp. Fig. 3.3 Polymorphism revealed by ISSR (R02) markers distinguishes between K7 and CRB at the fragment length of 420 bp. Fig 3.4 Polymorphisms among varieties revealed by SSR markers (CM16) 58 Fig. 3.5 Band frequency, private bands and mean heterozygosity across K7, CRB and Catuai. Fig. 3.6 Molecular Variance based on the calculation of PhiPT (ΦPT ) for the SSR data of 76 genotypes (K7: 54 and CRB: 22). Fig. 3.7 Phylogenetic tree constructed by dissimilarity matrix (Nei 72) and Neighbour Joining method (NTSYSpc 2.1) using SSR data. Fig. 3.8 Polymorphisms among varieties revealed by AFLP markers (EcoRI-AAC and MseI-CAA combination). Fig. 3.9 Pair-wise comparison of genetic dissimilarity among individuals between K7 and CRB (76 individuals) using combined data. Fig Polymorphisms within K7 variety revealed by AFLP markers (EcoRI- AAC and MseI-CAA combination). Fig Pair-wise comparison of genetic dissimilarity (Nei 72) within K7 in different blocks using combined data

11 Fig 3.12 Average genetic dissimilarity (Nei 72) within each cultivar using combined data. Fig Pair-wise comparison of genetic dissimilarity (Nei 72) within K7, CRB and Catuai using combined data. Fig Genetic relationship among K7 and CRB (76 individuals) revealed by Principle Coordinate Analysis (PCA) using combined data. Fig Genetic relationship among 95 individuals representing 7 varieties revealed by Principle Coordinate Analysis (PCA) using combined data. Fig Dendrogram generated from genetic distance using NJ method using combined data Table 2.1 RAPD primers used in the study. 29 Table 2.2 Reagent and its quantity used in restriction enzyme digests. 32 Table 2.3 RAPD PCRs. 33 Table 2.4 Reagents in an ISSR PCR. 34 Table 2.5 SSR primer information. 36 Table 2.6 SSR PCR. 36 Table 2.7 Digestion and ligation reaction in AFLP technique. 39 Table 2.8 PCR program for selective amplification step in AFLP technique. 40 Table 3.1 RAPD amplification results. 50 Table 3.2 ISSR primers and amplifications. 53 Table 3.3 Primer information and the results of SSR amplification. 54 Table 3.4 Genotypes of all accessions studied at four loci of microsatellites. 55 Table 3.5 Results of Analysis of Molecular Variance. 60 Table 3.6 AFLP combinations and the number of fragments amplified. 64 Table 3.7 DNA polymorphism of seven arabica varieties using SSR and AFLP data. 67

12 LIST OF ABBREVIATIONS AFLP CTAB DNA MW Marker ISSR NJ NNSW PCA PCR PVP RAPD RAPD/RE RE RFLP SEM SSR Ta Tm UPGMA VN WASI Amplified Fragment Length Polymorphism Cetyltrimethylammonium bromide DNA Molecular Weight Marker Inter-Simple Sequence Repeats Neighbour Joining Northern New South Wales Principle Coordinate Analysis Polymerase Chain Reaction Polyvinylpyrrolidone Randomly Amplified Polymorphic DNA RAPD primer : restriction enzyme combinations Restriction Enzyme Restriction Fragment Length Polymorphism Standard Error of the Means Simple Sequence Repeats (Microsatellites) Annealing temperature Melting temperature Unweighted Pair Group Method using Arithmetic Averages Vietnam Western Highland Agro-Forestry Science and Technical Institute

13 Arabica coffee plantation, Byron Bay, Australia Arabica coffee flowering, WASI, Vietnam Arabica coffee bearing fruit, WASI, Vietnam

14 CHAPTER I INTRODUCTION & LITERATURE REVIEW INTRODUCTION 1.1. Classification of coffee - taxonomy, origin and early history Coffee belongs to the genus Coffea, in the family Rubiaceae. There is a range of published taxonomies for this genus. According to Cramer (1957, cited in (Fazuoli et al. 2000), there are at least 100 species of coffee belonging to this genus, while Ferwerda (1976) classified 90 species of this genus. Based on the classification by Coste (1992), there are more than 70 species of coffee, of which two species, Coffea arabica L. (arabica) and Coffea canephora Pierre (robusta), are the main commercial species grown worldwide (Coste 1992). C. arabica coffee, which is indigenous to Ethiopia and comprises about 73% of world coffee production due to its superior quality (Orozco-Castillo et al. 1994) C. arabica utilisation has a longer history than C. canephora, probably more than 1,500 years ago and is now the most widespread species cultivated throughout the world (Coste 1992). C. arabica is predominantly autogamous selfing (Krug and Carvalho 1951; Graaff 1986; Clarke and Macrae 1988; Coste 1992; Wrigley 1995) and the only Coffea species that is tetraploid (2n = 4x = 44) (Krug and Carvalho 1951; Ferwerda 1976; Charrier and Berthaud 1985; Wrigley 1995). C. arabica has a very narrow genetic base (Ferwerda 1976; Vossen 1985) which will be described in detail. There were two genetic origins for this species, both of which spread from Yemen and gave rise to most of the commercial C. arabica grown worldwide, and two distinct botanical varieties C. arabica var. arabica (usually called Typica) and C. arabica var. bourbon (usually called Bourbon) (Krug and Carvalho 1951). Historical data indicate that the Typica genetic base originated from a single plant from Indonesia which was subsequently cultivated in the Amsterdam botanical garden in the early 18 th century, around 1715 (Chevalier and Dagron 1928, cited in Anthony et al. 2002a; Carvalho 1946, cited in Anthony et al. 2002a). The Bourbon genetic base 1

15 originated from a few coffee trees introduced from Mocha (Yemen) to the Bourbon Island (now La Reunion) at about the same time (Fig. 1.1) (Ferwerda 1976). Fig. 1.1: Schematic representation of the main steps in the history of coffee cultivation (after Anthony et al. 2002a). The narrow geographic origin of C. arabica, along with its self-fertilising nature and the historical or selective bottlenecks in its agricultural adoption, have resulted in low genetic diversity of C. arabica varieties cultivated around the world (Etienne et al. 2002; Martellossi et al. 2002; Steiger et al. 2002; Chaparro et al. 2004). This situation poses a challenge for identifying appropriate markers for cultivar identification based on DNA polymorphisms (see below, section 1.4). 2

16 1.2. Importance of coffee to agriculture Coffee is a popular beverage and hence a valuable agricultural export for its major producers. It is an important source of annual income and employment contributing significantly to the economies of developing countries within Africa, Asia and Latin America (Orozco-Castillo et al. 1994; Anthony et al. 2001). Presently there are approximately 6¾ million hectares worldwide under coffee plantation (Perfecto and Armbrecht 2003) producing around four million tons of green bean annually (Sera et al. 2000) with sales between 6 to 12 US billion dollars (Viniegra-Gonzalez 2000). The top ten coffee producing and exporting countries are led by Brazil and Vietnam, followed by Columbia, Indonesia, Mexico, Ivory Coast, Ethiopia, Uganda, Guatemala and India (Sera et al. 2000) Coffee in Northern New South Wales (NNSW) Australia Coffee is a fledging and expanding industry in Australia. The crop has been grown in Australia since 1890s and 1920s in Queensland and northern New South Wales, respectively, but declined significantly in the mid-20 th century (Lines-Kelly 1997). The current coffee growing areas are mainly in Far North Queensland and North-Eastern New South Wales which account for 600 ha of the total (740 ha) in Australia (Drinnan and Peasley 2004). The yield of C. arabica among 19 varieties grown in NNSW was between 3,000 kg (cultivar K7) and 1,500 kg (cultivar Caturra Rojo) green bean per hectare (Lines- Kelly 1997). The sub-tropical areas of Australia are suitable for growing coffee. However, the Australian coffee industry competes with the low cost of production in developing countries in the marketplace. It is thus vital for Australia to establish a clear defined niche marketing approach to supply to markets where the consumers have high standards of living and concern about quality, such as Japan and Germany (Jong 2000). Besides the marketing strategies, the fundamental factors to the industry will be product quality and consistency. Genetic consistency within varieties is the key to quality assurance in any agricultural product and is essential to projecting a high-quality niche image in the marketplace, particularly in high value crop products such as coffee. Furthermore, consistency of plant performance will greatly enhance efficiency of crop management practices. In 2000, a project was planned to extend the total area of coffee in Northern River Regions of NSW up to 1,253 ha by the year 2010 (Jong 2000). The freedom from serious and 3

17 widespread coffee diseases and pests suggests that pesticide-free or organically grown may become attractive advantages to quality of Australian arabica coffee (Drinnan and Peasley 2004). This quality, along with the consistency of the accessions grown, is the key to the success of the coffee industry Biological questions and hypothesis Morphological diversity vs single variety Molecular markers have revealed the loss of genetic diversity due to the history of C. arabica cultivation, the self-pollination of coffee and the selection by farmers (Lashermes et al. 1996b; Etienne et al. 2002; Anthony et al. 2002a; Masumbuko et al. 2003) (section 1.6.3). C. arabica accessions grown in NNSW are supposedly uniform cultivars. However, there is observed morphological variation in leaf shape, leaf size, leaf colour and tree height and shape within each arabica coffee variety in NNSW coffee plantations (Fig. 1.2 & Fig. 1.3). The literature showed very little genetic variation in cultivated C. arabica revealed by various molecular marker systems (section 1.6.3), which is contrary to the observed morphological variation within varieties of seed-propagated coffee in NNSW plantations. The level and sources of this morphological variation within and between coffee plantations of specific varieties are unknown and thus need to be determined. 4

18 Fig. 1.2 Off-type K7 on plantation, Byron Bay, Australia Fig. 1.3 True-to-type (left) and off-type (right) leaves of K7 5

19 Objective of the present study It is believed that morphological variability in coffee plantations is adverse to the product quality. While the Australian coffee industry targets high quality coffee, the morphological variability could lead to a lower value. Therefore there is a need to determine the causes of morphological variability within and between varieties in coffee. The morphological variability here may result from ecological factors, genetics or/and management factors such as mis-labelling or mis-planting. Due to the limitation of time, the ecological tests such as soil, sunlight and wind were not examined in this study. However, because the plants observed to be morphologically variable were within a limited geographic area with nearly homogenous soil type, ecological contribution to this variation is likely to be minimal. The genetic factors can be tested by molecular markers which can also shed light on the questions concerning crop quality management. Genetic variability could be detected by using molecular markers. The literature showed that there was low genetic variation or low polymorphism in arabica coffee (section 1.6.3). Hence a number of different molecular markers were used in this study to detect possible genetic variation of local coffee varieties on the Far North Coast region of NNSW and to perform DNA profiling of individual coffee plants within contiguous plantation blocks, among blocks of similar varieties and among blocks of different varieties. Having determined the extent of the genetic variability, an appropriate strategy could be developed which would reduce genetic variation for the long term benefit of the industry. The molecular technologies that can be used to address these genetic consistency issues are also applicable to other important questions for the world s coffee industries. For example, genetic improvement has been achieved by the development and application of traitspecific molecular markers to characteristics such things as resistance to coffee berry disease in arabica coffee (Agwanda et al. 1997), leaf rust resistance by using introgression molecular analysis (Prakash et al. 2004), the introgression of leaf rust and nematode resistance genes to C. arabica (Bertrand et al. 2003) and of biochemical compounds implied in coffee cupping quality (Barre et al. 1998). This study involved the determination of the most appropriate markers and their application to a range of appropriately selected plant samples. Furthermore, some Vietnamese C. arabica accessions collected from WASI (Western Highlands Agro- 6

20 Forestry Science and Technical Institute), a coffee research institute in Vietnam. This is a credited and reliable source which corresponds to those varieties in NNSW. Therefore, they were also included in the study to confirm the reliability of varietal selections grown and selected in NNSW and to compare the genetic variation among different origins. Therefore the purposes of the project were: - To assess the degree of genetic uniformity within and among C. arabica blocks using molecular markers. The results could feasibly inform the causes of morphological variability in NNSW coffee plantations. - To determine the most informative molecular marker systems for detecting genetic variation between the closely related varieties used in this study. Appropriate marker tools resulting from this study would be applied to varietal confirmation, trait-specific molecular markers and breeding strategies. - To compare the genetic profiles of Australian arabica coffee cultivars with reference cultivar types from Vietnam. The results would imply the need for a more stringent system to guarantee genetic purity of variety. LITERATURE REVIEW 1.5. Coffee breeding Coffee propagation For C. canephora, a self-incompatible coffee species, vegetative propagation methods are preferred and currently used for multiplying selected varieties of this species. In comparison, most commercially grown C. arabica cultivars have been propagated by seed as a standard practice since it is generally believed that arabica coffee is sufficiently truebreeding (Clifford and Willson 1985; Vossen 1985). Vegetative propagation methods are applicable to coffee, including cuttings, grafting and tissue culture. Propagation by cuttings is applied when few genotypes need to be propagated in large numbers. Grafting is a preferred method when a small number of plants are needed from each genotype. The advantage of grafting over cuttings is the vigour given to the scion by the rootstock. It is therefore possible to use grafting, as a research tool, to rescue weak seedlings or even haploid embryos (Clarke and Macrae 1988), or to produce the grafting with strong root systems that can be drought tolerant or disease resistant (Bittenbender et al. 2001). 7

21 In vitro methods can also be used for propagation by two ways: micro-cutting or somatic embryogenesis. This multiplication approach is able to produce a great number of plants but has the limitation of requiring refined techniques and chemical media (Clarke and Macrae 1988) Out-crossing in C. arabica Flower morphology in C. arabica makes selfing possible. However, there is a small percentage of cross-pollination by insects and wind. Furthermore, heterosis resulting from cross-fertilization allows for selection of superior phenotypes (Charrier and Berthaud 1985; Clifford and Willson 1985; Carvalho 1988). The first published data quantifying the percentage of natural cross-pollination in C. arabica, using the recessive gene marker prpr (for endosperm colour) was carried out by Taschdjian in 1932 (cited in Carvalho 1988) at Campinas (Brazil). The results from this study indicated a high percentage (about 50%) of cross-pollination. However a later study carried out in Campinas (Carvalho and Krug 1949) using an alternative recessive gene marker for endosperm colour indicated that the average percentage of seed produced from cross-fertilization was no greater than 9.0%. In experiments with other recessive markers (Carvalho 1962, cited in Carvalho 1988), cross-pollination rate ranges between 1.6% and 15.0%. A study in Jimma, Ethiopia (Meyer 1965) also detected a high level (40-60%) of out-crossing. These mixed results indicate that, although the nature of C. arabica is predominantly self-pollinating, the percentage of natural cross-pollination in this species is high enough to maintain a certain level of heterozygosity in the population (Carvalho 1988). These studies were mostly performed on wild populations where there existed diverse pollen sources, unlike the situation in plantations Mutants in C. arabica C. arabica mutants are very numerous and include variation in leaf shape and colour, growth habit, as well as flower, fruit and seed characters (Wrigley 1988). According to Clifford and Willson (1985), many current coffee cultivars are generally believed to be the consequence of spontaneous mutations of major genes influencing plant, fruit and seed characters, rather than the expression of residual heterozygosity. 8

22 More gene mutations have been found in the tetraploid C. arabica than in diploid C. canephora. This is probably because C. canephora is cross-pollinated, whereas C. arabica is largely self-fertilized, so there is more chance for recessive mutants to persist in the homozygous form in C. arabica, whereby they are expressed in the phenotype. In addition, a considerably greater number of plants of C. arabica are grown in the world, and they have been more extensively studied (Wrigley 1988). Lashermes et al (1999) stated that, due to its allotetraploid origin, C. arabica has a high level of fixed heterozygosity and therefore the level of internal genetic variability is twice that present in its diploid relatives. The authors contend that this variability may account for the success of C. arabica as a selected agricultural plant species. However, this apparently contradicts the evidence that C. arabica has been subjected to the selective bottleneck referred to earlier. Krug and Carvalho (1951) studied more than 40 mutants found in arabica coffee and contributed to a much better understanding of its genetics. They clearly demonstrated the diploid mode of inheritance of all characters in this allotetraploid (Vossen 1985) Biotechnology for coffee improvement Coffee, a perennial tree crop, is difficult to improve through traditional plant breeding. Conventional coffee breeding methodology faces considerable difficulties due to limitations such as the long generation time of coffee trees, the high cost of field trials and a lack of accuracy in the current breeding strategy (Etienne et al. 2002). For coffee, the tree only comes into bearing four or five years after the cross is carried out. The trees take over 10 years to reach their maximum production level and during these years the annual yields vary considerably. Therefore the identification of the elite mother trees requires a number of years of records and the progeny need to be replicated (Wrigley 1988). Recent advances in molecular genetics, when combined with conventional breeding, represent powerful tools for genetic improvement of coffee (Sondahl and Loh 1988; Wrigley 1988). The development of various biotechnological approaches such as micro- 9

23 propagation techniques, embryo rescue, anther culture, molecular markers and Marker Assisted Selection (MAS), genetic map development and genetic transformation have tremendous potential for genetic improvement of coffee (Lashermes et al. 2000c; Etienne et al. 2002). These techniques could overcome the limitations of traditional plant breeding and greatly enhance the efforts of coffee breeders (Sondahl and Loh 1988; Wrigley 1988; Etienne et al. 2002; Chaparro et al. 2004). Molecular analysis of C. arabica cultivars could provide knowledge of the levels of genetic variation and the genetic relatedness between genotypes which can improve the efficiency of utilisation of current germplasm resources. Furthermore, genetic data are important for designing effective plant breeding programs, by influencing the choice of parental genotypes to cross for the development of new populations (Russell et al. 1997) 1.6. Genetic markers for coffee Despite the enormous economic importance of C. arabica to coffee producers, there has been limited research into the genetic diversity in this species and its cultivated varieties. To date, detecting genetic variation has been hampered due to C. arabica s limited genetic origins and self pollination, compounded with historical and selective bottlenecks. The application of molecular techniques could greatly enhance future coffee genetic improvement programs (Etienne et al. 2002; Martellossi et al. 2002; Steiger et al. 2002; Chaparro et al. 2004). Knowledge of the levels and patterns of genetic diversity in C. arabica cultivars is required. Molecular techniques have been used in genetic studies of coffee including genetic diversity, identification of introgression fragments, genetic map development and cloning of economically important coffee genes (Etienne et al. 2002) Morphological markers Morphological markers are a classical method to distinguish variation based on the observation of the external morphological differences such as the size and shape of the leaf and of the plant form, the colour of the shoot tip, the characteristics of the fruit, the angle of branching and the length of the internodes. However, assessing polymorphism with morphological markers can be inefficient and they are generally dominant traits. Moreover, they often exhibit epistatic interactions with other genetic traits and can also be influenced by the environment (Vienne 2003). 10

24 There are marked morphological differences between the two main botanical types of C. arabica. Typica is considered to be the primitive form and has vigorous, pendulous primary bearing branches with horizontal habit, narrow leaves and bronze terminal leaflets while Bourbon has entirely green leaves and the fruit-bearing branches bent down only at the tips (Rothfos 1985; Wrigley 1988; Laak 1992) (Fig. 1.4 & Fig. 1.5). Entire green leaves Bronze terminal leaflets Fig. 1.4 Typical leaf colour of CRB and K7. 11

25 CRB (Bourbon type) with upright habit K7 (Typica type) with horizontal habit branches Fig. 1.5 Branching characters of CRB and K Biochemical markers Isozymes Enzymes are the basic tools of cellular chemistry and were introduced as genetic markers in the early 1970s (Glaubitz and Moran 2000). Many enzymes can exist in several forms, called isozymes, with slightly different molecular structures, arising from different alleles. These forms migrate at different rates when placed in a suitable medium within an electric field because they possess different electric charges or sizes. After migration, enzymes are provided with their specific substrate and the enzymatic reaction is coupled with a colour change reagent. It is then possible to observe coloured bands corresponding to isozymes. Isozyme analyses can be carried out using different plant tissues. This method has increased sensitivity for detecting variation over the classic morphological markers and is not as affected by environment (Berthaud and Charrier 1988). Isozyme markers have several limitations such as low polymorphism, some dependence on growth stage and conditions of plant growth and variation in the presence or activity of the enzymes in different tissues (Godwin et al. 2001; Vienne 2003). 12

26 Analysis of isozymes specifically in C. arabica accessions failed to reveal the amount of polymorphism detected using morphological markers, suggesting that isozyme markers are inappropriate for determining the genetic diversity in C. arabica (Berthaud and Charrier 1988; Lopes 1993; Bustamanate and Polanco 1999) DNA-based molecular markers A variety of molecular techniques available for detecting genetic diversity in coffee has opened up new possibilities for genetic analysis and provides new tools for the efficient use of coffee genetic resources (Etienne et al. 2002) Hybridisation-based (non-pcr) markers RFLP markers Restriction Fragment Length Polymorphism (RFLP) was first used as a tool for genetic studies in 1974 (Botstein et al. 1980). RFLP markers are based upon hybridization of a probe, a specific DNA sequence designed to hybridize with genomic DNA and detect a target sequence(s) in the unknown sample, on filters containing genomic DNA which has been digested with restriction enzymes. Fragments produced by restriction enzyme digestion are separated by gel electrophoresis and transferred onto filters by Southern blotting. Differences in the sequence at or around the sequence to which the probe hybridizes may result in differences (polymorphism) in length of the fragment detected by the probe (Grant and Shoemaker 1997; Henry 1997). RFLP is a codominant marker (allelic information) which has several advantages such as no sequence information required, unlimited number of loci and robustness in usage. However, this technique requires a relatively large amount of DNA in a form that is able to be digested with the restriction enzymes, is labour intensive and fairly expensive, displays low levels of polymorphism and is time consuming (Grant and Shoemaker 1997; Gupta et al. 1999). RFLP markers in coffee The first application of molecular genetics to coffee improvement was in the development of DNA probes as molecular fingerprints for protection of proprietary varieties. The same probes here were suggested as genetic markers to aid breeding (Sondahl and Loh 1988). 13

27 Unlike isozyme markers, genetic analysis using RFLP markers is not subject to environmental effects, nor is it confined to a specific stage of plant development. RFLP markers are thus completely neutral in certain circumstances (Sondahl and Loh 1988). However, establishment of genetic maps for coffee based on isozymes and RFLPs, either from total or chloroplast DNA, was not successful due to the low level of polymorphism (Lashermes et al. 1996a). Lashermes et al. (1999) used RFLP markers to investigate the origin and the diversity of C. arabica using twenty-three probes and revealed extremely low polymorphism among arabica accessions. The proportion of polymorphic loci among three accessions of C. arabica was only 7% PCR-based markers The Polymerase Chain Reaction (PCR) has become essential in studies of molecular ecology and population genetics research in the brief time since its invention. With this technique, defined DNA segments can be amplified to microgram quantities from as little as a single template molecule (Hoelzel and Green 1998). A wide variety of PCR-based marker techniques has been developed during the last decade, each with various advantages and different shortcomings (McGregor et al. 2000). These PCR-based markers differ in principle, application, complexity, informativeness, cost and time requirements (Mignouna et al. 2003). The choice of marker system often depends on the crop investigated (Powell et al. 1996a; Milbourne et al. 1997; Russell et al. 1997). The present study explored RAPD, ISSR, SSR and AFLP marker techniques. RAPD markers Randomly Amplified Polymorphic DNA (RAPD) is a PCR-based marker system, jointly described by Williams et al (1990) and Welsh and McClelland (1990). Amplification of genomic DNA using single primers of arbitrary nucleotide sequence, in low stringency conditions, results in multiple amplification products from loci distributed throughout the genome (Welsh and McClelland 1990; Williams et al. 1990). RAPD markers became popular because of their simplicity, applicability to any genome, no sequence information requirement, relatively small DNA quantities required, results 14

28 obtained quickly and high genomic abundance. Despite this, the limitations of RAPDs are numerous and include: they are dominant markers (i.e. cannot distinguish homozygotes from heterozygotes), are sensitive to laboratory changes and have low reproducibility within and between laboratories (Rafalski 1997). The number and pattern of bands amplified can be affected with variation in template concentration and with annealing, extension and denaturing time (Bielawski and Noack 1995) and template quality (Micheli et al. 1994). The original RAPD marker analysis has been modified by using restriction enzymes to cut the genomic DNA before the amplification. Riede and et al. (1994) and Koebner (1995) showed that in wheat, restriction digestion of genomic DNA with different endonucleases before PCR amplification improved the level of polymorphism and reduced non-specific amplification. The results were probably due to a greater efficiency in primer annealing along shorter DNA fragments where a simplified secondary DNA structure is less likely to interfere with the process. RAPD markers in coffee RAPD makers have been used to confirm the relationships within the genus Coffea (Orozco-Castillo et al. 1996), to construct a linkage map in coffee (Paillard et al. 1996; Lashermes et al. 1996a), to detect markers for resistance to coffee berry disease (Agwanda et al. 1997) and coffee leaf rust (Rani et al. 2000), and to study genetic diversity amongst wild accessions (Orozco-Castillo et al. 1994; Lashermes et al. 1996b; Anthony et al. 2001; Aga et al. 2003; Silveira et al. 2003; Chaparro et al. 2004) and cultivated varieties (Lashermes et al. 1996b; Masumbuko et al. 2003; Sera et al. 2003; Crochemore et al. 2004). Almost all studies confirmed the low polymorphism observed in the species C. arabica at the DNA level, even with wild populations. There is only one study that showed a moderate level of polymorphism (24 out of 42 primers) detected among wild and semiwild C. arabica accessions from Ethiopia (Chaparro et al. 2004). For cultivated accessions, Lashermes et al. (1996b) employed RAPD markers to study genetic diversity between cultivated and wild accessions of C. arabica and found the RAPD method appeared to be effective in resolving genetic variation and in grouping 15

29 germplasm in C. arabica. The study also confirmed the narrow genetic base of commercial cultivars of C. arabica, and only 12 of 140 RAPD primers detected polymorphisms and were not able to distinguish the cultivars within the same type groups, either Bourbon or Typica. RAPD markers were used to investigate the genetic diversity in Tanzanian cultivated C. arabica and found variability in these accessions and that these accessions clustered according to geographical locations (Masumbuko et al. 2003). However, only ten out of 100 decamer primers exhibited polymorphism and gave reproducible banding patterns. Similarly, Crochemore et al. (2004) employed RAPD markers to identify lines and cultivars of C. arabica of commercial interest. Four out of 56 primers showed polymorphism, but this low level of polymorphism suggested that RAPD markers did not have the capacity to consistently distinguish C. arabica cultivars. Agwanda et al. (1997) used 286 random decamer primers to detect the markers for resistance to coffee berry disease, Colletotrichum kahawae, in arabica coffee. The study also implied a narrow genetic base of cultivated C. arabica varieties. The RAPD technique associated with a prior digestion of genomic DNA with restriction endonucleases was applied to 14 elite Coffea arabica cultivars in Brazil in order to estimate their genetic variability (Sera et al. 2003) and to assess genetic variability within and among nine populations of C. arabica (Silveira et al. 2003). The considerable polymorphism detected in these studies illustrated that it is possible to find genetic divergence among C. arabica accessions of the same origin and the modified RAPD technique seems to be a useful tool for genetic characterization of C. arabica. Thus far, RAPD markers appear not to be very effective in revealing the genetic variation in C. arabica especially in cultivated populations. However, the modified RAPD technique - RAPD associated with a prior digestion of genomic DNA with restriction endonucleases - offered promise and therefore warrants further investigation. SSR markers Microsatellites are also known as simple sequence repeats (SSRs) and consist of segments of DNA containing tandem repeats typically of simple motif sequences of one to six bases (Litt and Luty 1989; Beckman and Weber 1992; Gupta et al. 1999; Hancock 1999; Glaubitz and Moran 2000). These microsatellite repeats are often flanked by unique 16

30 sequences, occurring only once in the genome (Glaubitz and Moran 2000). Microsatellites were first developed for use in the human genome (Weber and May 1989) and were later found to be abundant in plants (Morgante and Olivieri 1993). They are useful because they are highly polymorphic markers, i.e. they often have many alleles at each locus with each allele having a different number of tandem repeats (Beckman and Weber 1992) and length polymorphism can be revealed by electrophoresis. Microsatellites have been particularly useful for comparative genetics and genomic mapping since microsatellite sequences show a high degree of length polymorphism, abundance and distribution throughout the genome (Oh and Mao 1999; Glaubitz and Moran 2000). The relative abundance, multiallelic nature, codominant inheritance, high reproducibility and good genome coverage of microsatellites make them powerful molecular markers for use in genetic studies (Weber and May 1989; Morgante and Olivieri 1993; Powell et al. 1996a; Powell et al. 1996b; Glaubitz and Moran 2000). Also, SSR markers are very robust and user-friendly when compared with other marker systems and can be multiplexed during PCR or gel loading (Godwin et al. 2001). Furthermore, once SSR markers have been developed for use in one species it may be possible to use them in related species since the flanking regions are rather conserved and the repeats are relatively variable. This approach has been employed in grape (Arnold et al. 2002; Rossetto et al. 2002), sawfly (Hartel et al. 2003), yam (Hochu et al. 2005), hazelnut (Boccacci et al. 2005), bamboo (Nayak and Rout 2005). The greatest challenging using SSR is their discovery and initial development requiring species-specific DNA information for each marker locus in order to design primers from flanking regions (Godwin et al. 2001). This is impractical for many plant and animal species that do not already have well characterized genetic systems and for which the resources needed to carry out a labour-intensive SSR cloning and sequencing effort are not available (Henry 1997; Vogel and Scolnik 1997). There have been reports questioning the fidelity of microsatellite analysis due to electrophoresis artefacts (Fernando et al. 2001), low extension temperatures (Hite et al. 1996) and possible null alleles (Donini et al. 1998; Yazdani et al. 2003). SSR markers in coffee Microsatellites were applied in C. arabica to identify C. arabica, C. canephora and related coffee species (Combes et al. 2000), to investigate polymorphisms among wild and cultivated C. arabica accessions (Rovelli et al. 2000; Anthony et al. 2002b; Baruah et al. 17

31 2003; Moncada and Couch 2004), to evaluate the quality of green bean, roasted beans and instant coffee (Martellossi et al. 2002; Taylor et al. 2002; Palmieri et al. 2003) and to study gene introgression (Herrera et al. 2002b). The microsatellites also showed low genetic variation in C. arabica compared to that detected by RAPDs. Combes et al. (2000) used 11 SSR primers to characterize microsatellite loci in 32 individuals of C. arabica, 10 individuals of C. canephora and related coffee species, representing a total of 13 Coffea taxa. Only five out of 11 microsatellite loci appeared to be polymorphic in C. arabica and the mean heterozygosity value was ten times lower than C. canephora (0.04 and 0.47, respectively). Although the primers were designed from sequences isolated in C. arabica, they worked well for most of the diploid coffee species Combes et al. (2000). These microsatellite primers, however, found little variation in C. arabica. Only five out of 11 primer pairs revealed polymorphisms in C. arabica, probably as the consequence of its origin, reproductive biology and evolution (Lashermes et al. 1999). Similarly, nine SSR markers were developed by Baruah et al. (2003) to identify polymorphism in C. arabica, C. canephora and 17 species of Coffea and the related genera Pilanthus also revealed very low polymorphism across the 45 arabica genotypes. The SSR primers of Rovelli et al. (2000), isolated from two genomic libraries of C. arabica, showed polymorphisms among the C. arabica accessions studied and were able to discriminate between the two chromosome sets derived from the diploid donor ancestral plants. Thirty-one SSR primer pairs were used to assess polymorphism among 16 C. arabica and four C. canephora accessions and to identify DNA introgression fragments from C. canephora in four C. arabica lines (Anthony et al. 2000). These primers were able to distinguish C. arabica accessions into two groups according to their genetic origin. Moncada and Couch (2004) used thirty-four fluorescently labelled microsatellite markers to assess genetic diversity among five diploid species and 23 various cultivated and wild accessions of tetraploid C. arabica from Colombia The results showed higher genetic polymorphism in comparison with previous reports. The primer pairs used could detect variation among accessions with a very high percentage of polymorphism (56%, calculated by the percentage of markers detecting variation), even within cultivated C. arabica. Eighteen percent of SSR alleles showed intra-subspecific polymorphism. 18

32 SSR markers have also been applied to evaluate the quality of green bean, roasted beans and instant coffee in order to evaluate quality, as a diagnostic tool for food authenticity, provenance and traceability of variety composition of complex food matrices (Martellossi et al. 2002; Taylor et al. 2002; Palmieri et al. 2003). In these studies, SSRs were able to distinguish between C. arabica and C. canephora species. The combined use of SSR and RFLP or AFLP markers were useful in study the gene introgression into C. arabica by way of triploid hybrids (C. arabica x C. canephora) (Herrera et al. 2002b) and by way of tetraploid interspecific hybrids (C. arabica x C. canephora) (Herrera et al. 2002a). The results found that SSRs and AFLPs were appropriate markers for studying introgression in coffee. The number of individuals exhibiting C. canephora-specific markers among backcrosses of F 1 populations (BC 1 ) was rather small for the triploid interspecific hybrids. However a significantly higher frequency of C. canephora-specific markers was revealed in BC 1 populations for tetraploid interspecific hybrids. Overall, SSR markers showed little variation in C. arabica. However, because the most recent study of Moncada and Couch (2004) suggested the potential of SSR markers in the study of genetic diversity, SSRs were utilised in the study reported here. ISSR markers The Inter-SSR (ISSR) also called MP-PCR (microsatellite-primed PCR) or ISA (Inter-SSR Amplification) or RAMP (Randomly Amplified Microsatellite Amplification) marker system was first developed by Zietkiewicz et al. (1994). The ISSR technique is a PCR based method, which involves amplification of DNA segments present at an amplifiable distance in between two identical microsatellite repeat regions oriented in opposite direction (Reddy et al. 2002). Inter-SSRs are semi-arbitrary markers amplified by PCR in the presence of one or two primers complementary to a target microsatellite. Similar to RAPD markers, ISSR markers do not require genome sequence information but provide multi-loci amplification (Bornet and Branchard 2001). There are two types of SSR-directed (or ISSR) primers: unanchored and anchored. Unanchored ISSR markers rely on the presence of two microsatellites having the same repeat unit in an inverse orientation and separated by an amplifiable distance within the 19

33 genome, so that the inter-repeat sequences (ISSRs) are amplified (Gupta et al. 1999). According to Reddy et al. (2002), when unanchored, the primer tends to slip within the repeat unit during amplification leading to smears instead of clear bands. Anchored primers will include not only the repeat sequence but also two to four non-repeat degenerate nucleotides extended at either 3 or 5 ends (Zietkiewicz et al. 1994). An anchored primer assures annealing only to the ends of a microsatellite in template DNA thus obviating internal priming and smear formation (Reddy et al. 2002). ISSR markers are as quick and easy to handle as RAPD markers, but their reproducibility is still uncertain. Some reports showed that ISSR markers are reliable in wheat (Nagaoka and Ogihara 1997) and in cotton (Liu and Wendel 2001) and may have the reproducibility of SSR markers because of the longer primer length (Bornet and Branchard 2001). This means high annealing temperature applied can result in higher stringency (Gupta et al. 1999). However, several studies have indicated that, despite the higher annealing conditions used, these amplifications behave similarly to RAPD in terms of susceptibility to slight changes in reaction conditions such as annealing temperature, primer concentration and magnesium concentration. All these changes affect the quality of the banding patterns (Weising et al. 1995). ISSR markers in coffee ISSR markers have been used in several studies in coffee. Rani et al. (2000) employed ISSR markers combined with other markers (RFLP, RAPD) to study the genetic variability and novel genome organizations in the commercially well-established somatic embryogenesis-derived plants of C. arabica. Seventeen of 25 repeat motifs used as primers were found to produce good analysable amplification products. Fourteen ISSR makers were also useful in evaluating genetic differentiation among eight Coffea species and determine the parentage of six interspecific hybrids (Ruas et al. 2003). Since the studies above demonstrated the capacity of ISSR markers to investigate genetic variation in coffee, ISSRs were employed in the present study. 20

34 AFLP markers Amplified Fragment Length Polymorphism (AFLP) is based on the selective amplification of sets of restriction fragments from genomic DNA (Vos et al. 1995). DNA is first cut with restriction enzymes and double-stranded adaptors are then ligated to the fragments. PCR primers designed to match the sequence of the adaptors are used to amplify the fragments. The number of fragments amplified is intentionally limited by including nucleotides at the 3 end of the primer that extend into the unique sequence of the DNA fragment (Hoelzel and Green 1998). The key feature of AFLP-PCR is its capacity for the simultaneous screening of many different DNA regions distributed throughout the genome (Mueller and Wolfenbarger 1999). The great advantages of AFLPs are the large number of loci uncovered in a single assay (Pillay and Myers 1999; Glaubitz and Moran 2000), the high reproducibility (Powell et al. 1996a; Milbourne et al. 1997; Russell et al. 1997; Breyne et al. 1999; Gupta et al. 1999; Pillay and Myers 1999), no sequence information of the DNA required (Pillay and Myers 1999; Vienne et al. 2003) and automation (Henry 1997; Ticknor et al. 2001; Campbell and Bernatchez 2004). Mueller and Wolfenbarger (1999) reported that AFLPs have been used to discriminate between closely related species that had been impossible to resolve with morphological or other molecular systematic characters. However, AFLPs are dominant markers, their laboratory techniques are slightly challenging and they have been patented (Zabeau and Vos 1993). Some reports mention the production of artifacts or amplification failure of a fragment of AFLP probably due to restriction-ligation step (Mueller and Wolfenbarger 1999). Limitations can be minimized by complete digestion which prevents later amplification of uncut fragments using high quality DNA and an excess of restriction enzymes or using high annealing temperatures (Jones et al. 1997; Arens et al. 1998; Winfield et al. 1998). A study testing reproducibility of AFLPs among European laboratories (Jones et al. 1997) showed that AFLP profiles provided extremely high reproducibility and the analyses of AFLP scores from duplicate test samples revealed average errors of 0-2% (Jones et al. 1997; Arens et al. 1998; Winfield et al. 1998). In general, the high reliability of AFLP markers could lead to the replacement of RAPD markers, and their relative user-friendliness may cause a partial replacement of other highresolution markers such as RFLPs. However, because of their largely dominant nature, AFLP markers are unlikely to out-compete codominant markers such as microsatellites. 21

35 Therefore, AFLPs and microsatellites are both likely to remain key molecular tools for the analysis of genetic variation (Mueller and Wolfenbarger 1999). AFLP markers in coffee AFLP markers were used to detect the introgression of C. canephora genetic material in C. arabica (Lashermes et al. 2000a), the introgression of C. liberica genetic material in C. arabica (Prakash et al. 2002), the introgression of C. canephora into C. arabica in beverage quality (Prakash et al. 2002; Bertrand et al. 2003), to construct a genetic linkage map in coffee (Lashermes et al. 2001; Pearl et al. 2004), to identify the origin of cultivated C. arabica (Anthony et al. 2002a) and to study genetic polymorphisms in C. arabica (Steiger et al. 2002; Anthony et al. 2002b). In agreement with other marker systems, AFLP also showed a low genetic diversity in C. arabica. Anthony et al. (2002a) used AFLP markers to assess polymorphism between and within Typica and Bourbon derived accessions and Yemen cultivars. AFLP markers were able to differentiate between these two types. However, the results confirmed the low genetic variation within groups. Sixty-one Coffea accessions composed of six arabica cultivars were analysed with six AFLP primer combinations and showed small differences at the DNA level between and within cultivars (Steiger et al. 2002). Overall, the literature showed little genetic variation detected in C. arabica as determined by almost all marker types, especially for cultivated accessions. Because the accessions in this study also belong to cultivars, it was necessary to apply different marker types in order to identify informative markers. The research strategy was to first use RAPD and ISSR markers due to their simplicity and low cost. If these marker types failed to detect polymorphisms among accessions, then microsatellites and AFLPs were to be employed. 22

36 CHAPTER II - METHODS 2.1. Sample collection Sampling in Northern New South Wales, Australia A total of 95 C. arabica accessions were used in the study, of which 84 came from the NNSW and 11 were from the Central Highland of Vietnam. The 84 NNSW samples included representatives of the 7 main cultivars which produce the highest yields of the 19 cultivars grown in NNSW (Peasley and Winston 1997). Of these 7 cultivated varieties, the most widely grown in Byron Bay, NNSW are K7 and CRB which produced the highest yields (over 2500 kg/ha) of high quality dry green bean (Peasley and Winston 1997). The origin of K7 grown in NNSW is ambiguous. According to Peasley and Winston (1997) K7 originated from East Africa. Several documented sources indicated that K7 is a Kent commercial variety selected in Kenya (Walyawo 1983, cited in Lashermes et al. 1996b; Agwanda et al. 1997), named after coffee industry aficionado Mr Kent (Laak 1992). However, Narasimhaswamy (1950, cited in Vossen 1985) and Rothfos (1985) suggested that K7 was a Kent variety developed from a single tree selection in India, probably a hybrid between Typica and an unknown type. K7 is high yielding and resistant to rust and coffee berry disease. K7 accessions used in this study came from three different blocks named K7 C (C stands for Chesterfield plantation), K7 K (K stands for King plantation) and K7 M (M stands for Myers plantation) in which 24, 20 and 10 individuals in each block were collected, respectively. The origin of CRB, a local selection of the Bourbon type, is also somewhat equivocal. The name, CRB, was given by David Peasley, a horticultural consultant in Murwillumbah, NNSW, and is an acronym for Condong Range Bourbon. CRB is the only local selection in five varietal trials from a farm in Murwillumbah in (Peasley and Winston 1997). However, at the same time, Joy Phelps and Joan Dibden (Wombah coffee growers) selected a line and named it Clarence River Bourbon (CRB), thus generating some confusion. It is possibly the same as Peasley s selection and both appear identical to the old Bourbon type which has been grown in NNSW for over 100 years (Peasley, pers. comm.). CRB has greater cold tolerance than the other varieties (Peasley and Winston 1997). The CRB accessions were collected from 2 blocks named CRB C from Chesterfield plantation 23

37 and CRB K from King plantation consisting of two and 20 individuals per block, respectively. Other cultivars widely grown in NNSW are: Mundo Novo derived from a Bourbon and Typica crossing (Steiger et al. 2002; Anthony et al. 2002a; Silveira et al. 2003) from Costa Rica, Catuai - a hybrid between Mundo Novo and Caturra (a Bourbon mutant) (Coste 1992; Laak 1992; Anthony et al. 2002a) from Costa Rica, SL14 and SL34, Kenya collection series (Rothfos 1985) from East Africa, and Arusha from Nambour ex PNG (Peasley and Winston 1997). One to two individuals per collection for the accessions of Catuai, SL14, SL34, Arusha and Mundo Novo were also analysed to test the reliability of marker systems in cases where no polymorphism was found within K7 or within CRB and between them (Appendix 1). All 84 NNSW individuals in this study originated from seed of an unknown number of progenitor plants, except K7 individuals grown in Chesterfield Block which comes from four selected mother plants. The sampling of accessions was random, except K7 in block A, Chesterfield plantation, where a high incidence of morphological differences were observed, the sampling was based on the presence of individuals which presented morphological differences compared to normal individuals in the same block Sampling in Daklak, Vietnam The WASI - The Western Highland Agro-Forestry Science and Technical Institute - a former coffee research institute in Daklak, Vietnam - holds a coffee germplasm collection from various sources all over the world. Eleven samples, which correspond to five out of seven varieties grown in NNSW were collected from WASI to use as references for the same C. arabica varieties from NNSW, Australia. The 11 individuals comprised four Catuai from South Africa, Costa Rica, Brazil and Cuba, one SL14 from Thailand, one Mundo Novo from Brazil, one K7 from South Africa and four Bourbon (to compare with CRB from Australia) from Cameroon and Vietnam (see Appendix 1) 24

38 2.2. DNA extraction Sample storage Fully expanded terminal coffee leaves collected from the field were placed into paper bags, kept at low temperature and then stored by several methods: 1. For immediate use: Analysis of fresh tissues is the preferred option for DNA extraction in most cases. Fresh leaf may be extracted immediately or stored for several days at cool temperature before extraction. 2. For use in 7-14 days: Samples are stored in refrigeration at 4 0 C. This type of storage usually allows extraction to be delayed for 1-2 weeks. Storage conditions that discourage the growth of fungi in these tissues are desirable. 3. Long term storage of dry or frozen tissue is possible for coffee leaves: For dry storage, coffee leaves were put into plastic bags with self-indicating silica gel. With frozen storage methods, degradation or disruption of DNA occurs upon thawing. The separation of samples into different small bags or tubes enables DNA extraction of sub-samples thereby avoiding repeated freeze/thawing Different methods of DNA extraction C. arabica is a plant species especially rich in polyphenolics (Labat et al. 2000). High amounts of polysaccharides, polyphenolics and secondary metabolites in leaves is detrimental to DNA extraction as reported in several plants such as cotton (Chaudhry et al. 1999), tomato (Peterson et al. 1997) and cacti (Tel-zur et al. 1999). These compounds, if present in the DNA extraction, inhibit PCR amplification (Bi et al. 1996; Ky et al. 2000; Steiger et al. 2002). Prior to studying the marker system to be used, optimisation of DNA extraction to ensure best quality was performed in the following manner. Initial DNA extraction employed the traditional cetyltrimethylammonium bromide (CTAB) method (Doyle and Doyle 1987) which was unable to obtain high yield of DNA due to the apparent presence of polysaccharides and secondary metabolites. The method of DNA extraction was then modified with the addition of 2% polyvinylpyrrolidone (PVP) in 3% CTAB buffer to bind polyphenols (Lodhi et al. 1994; Chaudhry et al. 1999; Khanuja et al. 1999) and using Phenol:Chloroform:Isoamyl alcohol 25:24:1 to remove protein. Using this method, DNA yield was higher than from the traditional method. Although the 25

39 absorbance of DNA at 260 was high ranging from 0.3 to 0.9 (approximately µg DNA/µl solution), the OD260/OD280 value was low (0.7), indicating the presence of contaminants such as polysaccharides and secondary metabolites. To test if the DNA was amplifiable, PCR was performed using CM16, a microsatellite primer pair that worked well on C. arabica (Baruah et al. 2003). The DNA solution from these extractions was brown and sticky and no DNA fragment was amplified by PCR (Fig. 2.1). MW Marker II CTAB method with modifications Method using activated charcoal Method using Qiagen kit Fig. 2.1 DNA amplification of PCR indicates the effect of different methods of DNA extraction DNA MW Marker II: Size marker ( kbp) (Cat. No , Roche Applied Science, Germany). Bi et al. (1996) used activated charcoal in their extraction method to obtain improved DNA quality in cotton, coffee, rubber tree, cassava and banana. The DNA solutions were clear 26

40 and amplified well with PCR due to the effectiveness of activated charcoal in removing coloured and resinous matter which inhibits DNA polymerase during amplification. Bi et al. (1996) modified the extraction procedure as follows: 30 mg of PVP, 100 mg of CTAB and 12.5 mg of activated charcoal added in the mortar in the presence of liquid nitrogen for grinding. Five percent β-mercaptoethanol working as antioxidant (Chaudhry et al. 1999) was added in 3% CTAB to remove the polyphenols and improved the purity of the DNA extract (Paterson et al. 1993; Bi et al. 1996; Khanuja et al. 1999). Another method of DNA extraction using Qiagen kit (DNeasy Plant Mini Kit (50)-Cat No 69104) produced good quality DNA, but this technique was not as cost effective as the other extraction methods used. DNA extraction protocol used in this study 250 mg of fresh leaves taken from the first or second mature youngest leaves of the C. arabica trees was ground in liquid nitrogen by mortar and pestle in the presence of 30 mg of PVP, 100 mg of CTAB and 12.5 mg of activated charcoal. The frozen powder was then transferred to two 1.5 ml tubes each containing 600 µl of 3% CTAB and 5% β- Mercaptoethanol, preheated to 65 0 C for 10 minutes in a water bath. The solution was incubated in a water bath at 65 0 C for minutes and was shaken 2-3 times during incubation. Following incubation, the solution was inverted several times after adding 600 µl of Chloroform:Isoamyl-alcohol 24:1 to bind protein and then centrifuged for 3 minutes at 10,000 rpm. The top layer was removed gently and put into a new tube with another 600 µl of Chloroform:Isoamyl-alcohol 24:1. The solution was inverted several times and centrifuged again as above. The top layer was removed into a clean tube and mixed well with 600 µl of ice-cold isopropanol and 75 µl of sodium acetate (3M, ph= ), then put into freezer at C for 30 minutes for the precipitation of DNA. The tube was then centrifuged at 14,000 rpm for 20 minutes to form a pellet. All supernatant was removed without disturbing the pellet which was then washed with 500 µl of 70% ethanol to remove the supernatant and any salts. The solution was centrifuged at 10,000 rpm for 3 minutes and the supernatant removed. The pellet was then left in the open air or dried in a vacuum dryer for 1-2 minutes to remove any residual ethanol. The pellet 27

41 was finally dissolved in 80µl of autoclaved MQ water or TE buffer ready to use. The DNA solution was stored frozen at C for long term usage. The DNA obtained was checked by both spectrophotometer measurement (NanoDrop ND1000 Full-spectrum UV/Vis Spectrophotometer - NanoDrop Technologies, Inc.) and agarose gel quantification. The DNA quantity ranged from 7.3 (ng/µl) to (ng/µl). The absorbance of DNA at 260 nm varied between and and A260/A280 was from 1.40 to There were 38 samples out of 95 with A260/A280 greater than 1.8. DNA concentration was standardised to ng/µl for use PCR optimisation In this study, four different DNA marker systems were used to evaluate the genetic diversity of local C. arabica cultivars collected from four different plantations in NNSW. The results provided an insight into the degree of genetic variation within seed-propagated plantation varieties, which may impact on coffee quality, the consistency of product and the convenience of plantation management RAPD markers RAPD markers, modified by restriction digestion of genomic DNA before PCR amplification (as described in Chapter 1), were used in this study (Sera et al. 2003; Silveira et al. 2003) Primers and enzymes A total of 20 PCR primers synthesized from Proligo Primers and Probes (Proligo Australia Pty Ltd) were used in this study (Table 2.1). The sequence information of these primers was obtained from various published coffee studies (Paillard et al. 1996; Lashermes et al. 1996b; Agwanda et al. 1997; Anthony et al. 2001; Aga et al. 2003; Masumbuko et al. 2003; Silveira et al. 2003; Chaparro et al. 2004). 28

42 Table 2.1 RAPD primers used in the study No Primers Sequence 5 to 3 No Primers Sequence 5 to 3 1 OPA07 GAAACGGGTG 11 OPN20 GGTGCTCCGT 2 OPB02 TGATCCCTGG 12 OPP18 GGCTTGGCCT 3 OPC07 GTCCCGACGA 13 OPS15 CAGTTCACGG 4 OPC10 TGTCTGGGGTG 14 OPX09 GGTCTGGTTG 5 OPC15 GACGGATCAG 15 OPZ07 CCAGGAGGAC 6 OPF05 CCGAATTCCC 16 OPZ14 TCGGAGGTTC 7 OPI20 AAAGTGCGGG 17 UBC 16 GGTGGCGGGA 8 OPK14 CCCGCTACAC 18 UBC 212 GCTGCGTGAC 9 OPK18 CCTAGTCGAG 19 UBC 217 ACAGGTAGAC 10 OPN18 GGTGAGGTCA 20 UBC 220 GTCGATGTCG Origin of primer sequence: OP = Operon, from Operon Technologies Inc. (Ca, USA) UBC = University of British Columbia, from University of British Columbia (Vancouver, Canada) A total of five restriction enzymes used in the study were: HindIII, EcoRI, BamHI, RsaI and HaeIII. (New England Biolabs Inc). Recognition site: HindIII 5..A/AGCTT.3 5..TTCGA/A 3 EcoRI 5..G/AATTC.3 3..CTTAA/G.5 BamHI 5..G/GATCC 3 3..CCTAG/G 5 RsaI 5..GT/AC.3 3..CA/TG.5 HaeIII 5..GG/CC.3 3..CC/GG.5 Two out of the five enzymes (RsaI and HaeIII) are frequent cutters and the remaining enzymes are rare cutters. The cutting capacity of each enzyme was then examined using agarose gel electrophoresis. As expected, the two frequent cutters digested the genomic 29

43 DNA well while the other three rare cutters digested less efficiently, especially EcoRI and BamHI (see Fig. 2.2). Fig. 2.2 Restriction efficiency of five enzymes conducted on four repeated samples. DNA MW Marker II: Size marker ( kbp) (Cat. No , Roche Applied Science, Germany). DNA MW Marker XIV: Size marker ( bp) (Cat. No , Roche Applied Science, Germany). No 1 - No 4: Sample names. 30

44 In order to investigate genetic polymorphisms in coffee samples by RAPD analysis, it was necessary to optimise the reaction conditions. Besides the primer and the quality of DNA template, the efficiency of PCR is controlled by many parameters such as MgCl 2 concentration, annealing temperature, dntp concentration, DNA polymerase, buffer type and cycling parameters (Innis and Gelfand 1999). For routine PCR, PE 9700 thermal cyclers were employed using the conditions described below MgCl 2 concentration High levels of non-specific amplification and low product yield can be the results of using suboptimal concentrations of MgCl 2 (Zangenberg et al. 1999). Several MgCl 2 concentrations (0.5, 1.0, 1.5, 2.0 and 2.5 mm) were used in this study to find the optimum. Both 1.5 and 2.0 mm concentration gave clear and bright bands on agarose gels. The concentration of 1.5 mm however was chosen as a standard concentration for PCR since this concentration is already included in Taq buffer provided by Roche Annealing temperature (Ta) The selection of appropriate annealing temperature depends on the melting temperature of the primers (Zangenberg et al. 1999). Generally, more specific amplification can be obtained by using higher annealing temperatures but the loss of longer target fragments and overall reduction of PCR product yield can occur. In contrast, lower temperatures often result in low reproducibility, but specific target amplification can be achieved (Zangenberg et al. 1999). To examine the suitable annealing temperature for each primer, PCRs were run on a temperature gradient PCR machine (Palm cycler, Corbett Research) which has a capability of providing up to 12 different annealing temperatures simultaneously across a series of equivalent samples in order to determine the optimum. The range of annealing temperatures assessed was 32 C - 43 C DNA template preparation All DNA templates were standardised to a concentration of approximately ng/µl, as determined by a Nanodrop ND1000 spectrophotometer. However, due to phenolic compounds in coffee DNA template, some samples were diluted further to reduce the adverse impact of contaminants on the amplification. 31

45 Firstly, genomic DNA was digested with one of five enzymes (HindIII, EcoRI, BamHI, RsaI and HaeIII) (20,000 U/ml for HindHI, EcoRI, BamHI and 10,000 U/ml for RsaI and HaeIII). The digestion reaction was carried out in a total volume of 25 µl and composed of 5 µl of template DNA, 200 U/ml of enzyme, 2.5 µl of enzyme buffer (10 x) and H 2 O to a total volume of 25 µl. For BamIII, a volume of 2.5 µl of BSA is required (Table 2.2). The mixture was then incubated in a water bath at 37 0 C overnight. An aliquot of 2.5 µl of the digested solution was used in subsequent PCRs, representing approximately ng of DNA template. The 25 µl working solution was sufficient for 10 PCRs for each sample. Table 2.2 Reagent and quantity used in restriction enzyme digests. Reagents Concentration Quantity (µl)* DNA ng/µl 5 BSA (only for Bam) 10 x 2.5 Buffer 10 x 2.5 Enzyme 10,000 (or ) U/ml 0.5 or 0.25 H 2 O Made up to 25 µl *Amount of reagent in 25 µl of reaction (µl) dntp concentration, DNA polymerase and buffer type dntps (Promega, Madison WI, Cat# U1240) were used at a concentration of 2.5 mm for each dntp. Taq DNA polymerase was used at 0.06U (Roche Diagnostics Australia Pty Ltd). PCRs were performed in Taq 10x buffer (Roche, provided with Taq and containing MgCl 2 at 1.5 mm final concentration) diluted to the standard concentration of 1 x. 32

46 PCR conditions The RAPD reactions were performed in a total volume of 25µl (Table 2.3) Table 2.3 RAPD PCRs. Reagents Stock solution Quantity (µl)* Final concentration Primer 20 µm µm Taq buffer (Roche) 10 x x Taq 5 U/µl U/µl dntps (each) 2.5 mm mm H 2 O DNA digested ng/µl ng/25 µl * Amount of reagent used in 25µl PCR reaction A master mix was prepared and PCR mixtures were positioned in the 96 wells. Amplifications were performed in an automated cycler (GeneAmp. PCR System 9700 PE Applied Biosystems) programmed for 20 min. at 85 o C or 65 o C as required to inactivate the restriction enzymes, followed by 5 min C denaturation, then 40 cycles of 30 sec. at 94 0 C, 105 sec. at from 32 0 C to 38 0 C (different temperatures depend on each primer), 60 sec. at 72 0 C followed by a final elongation of 10 min. at 72 0 C and remaining at 15 0 C. Initially, RAPD markers were used to assess the level of polymorphism in seven varieties. The assessment began with a small number of accessions including three samples for each of the two most predominant plantation varieties (K7 and CRB) and one sample for each of the five less common varieties in NNSW (Catuai, Arusha, SL14, SL34 and Mundo Novo). The primers which exhibited polymorphisms among these eleven samples were then tested against the remaining accessions (Appendix 1) ISSR markers Primers used in the study Ten single and 22 combinations of published coffee ISSR primers were selected (Ruas et al. 2003). The length of the primers ranged from 16 to 19 bases and contained different SSR dinucleotide (4 primers), trinucleotide (1 primer) or tetranucleotide repeat motifs (5 33

47 primers). Of these, two were anchored primers and eight unanchored primers. Primers were used either single or combined in pairs (see Table 3.2) PCR optimisation Except annealing temperature and cycling parameters, the concentrations of almost all parameters such as MgCl 2 concentration, dntp concentration, DNA polymerase and buffer type were similar to RAPD techniques. In order to determine the optimum annealing temperature for each primer, PCRs were run on the gradient PCR machine (Palm cycler, Corbett Research). The range of annealing temperatures depends on the melting temperature (Tm) of each primer (Fig 2.3). The lowest melting temperature was 40 0 C (primer R05) and the highest melting temperature was 58 0 C (primer R02). For routine PCR, PE 9700 thermal cyclers were employed using the conditions described below PCR The ISSR reactions were performed in a volume of 25µl and contained reagents and amounts as described in Table 2.4. Table 2.4 Reagents in an ISSR PCR Reagents Working solution Quantity (µl)* Final concentration Primer 20 µm µm Taq buffer (Roche) 10 x x Taq 5 U U/µl dntps (each) 2.5 mm mm H 2 O DNA template ng/µl ng/ 25µl * Amount of reagent used in 25µl PCR reaction A master mix was prepared and PCR mixtures were positioned in the 96 wells. Amplifications were performed in an automated cycler (GeneAmp PCR System 9700 PE Applied Biosystems) programmed for 5 min. of initial denaturation at 95 0 C, then 40 cycles of 30 sec. at 94 0 C, 60 sec. at from 48 0 C to 58 0 C (depending on primer), 60 sec. at 72 0 C as well as a final elongation cycle of 10 min. at 72 0 C and remaining at 15 0 C. For the combined primers, a touch-down annealing temperature regime was applied with a 34

48 decrease of C or 1 0 C for each cycle as required to achieve the final annealing temperature within 10 cycles (Fig. 2.3). -Initial denaturing: 94 0 C 5 min -Denaturing 94 0 C 1 min -Annealing: Ranging from 48 to 58 0 C 1 min -Extension: 72 0 C 1 min -Final extension: 72 0 C 10 min 40 cycles 44 0 C C: range of annealing temperatures Fig 2.3 PCR optimization to examine the annealing temperature using ISSR R02 primer SSR markers Primer information The six most informative SSR markers (Table 2.5) from previous studies (Baruah et al. 2003; Moncada and Couch 2004) were applied in this study. Fluorescently labelled primers were synthesized by Proligo (Hex and 6-Fam) and Applied Biosystems (Ned and Vic) in which fluorophores were added to the 5 end of the oligonucleotides for detection of SSR amplicons on the ABI capillary electrophoresis instruments (ABI Prism 310 Genetic Analyser and ABI 3730 DNA Analyser). Six primers (Table 2.5) were first run for 14 individuals from 7 varieties. 35

49 Table 2.5 SSR primer information No Marker Tm ( 0 C) Primer sequence (5-3 ) Source 1 CM2 -FAM 60 Forward: TGTGATGCCATTAGCCTAGC A 58 Reverse: TCCAACATGTGCTGGTGATT 2 CM8-FAM 56 Forward: GCCAATTGTGCAAAGTGCT A 58 Reverse: ATTCATGGGGCCTTTGTCTT 3 CM16-HEX 57 Forward: TGGGGAAAAGAAGGATATAGACAAGAG A 58 Reverse: GAGGGGGGCTAAGGGAATAACATA 4 CFGA Forward: CATCCATCCGAAAACTTGTAACG B NED 57 Reverse: CAGCACTGGCAAATAGCAACTCTT 5 CFGA Forward: AAGCCACCCAGAAAACAGCACATC B FAM 56 Reverse: ATTTGCTTCTCATGTTCCCTTTCA 6 CFGA547a- VIC Forward: AAGGCATGCGGCGGGAGTAT Reverse: TCGTCAAGGACAATCCTAAAGC B A Baruah et al. (2003) B Moncada and Couch (2004) PCR PCR was carried out using the amount of reagents as described in Table 2.6 Table 2.6 SSR PCR. Reagents Working solutions Volume for a PCR reaction (25 µl) Final concentration Forward primer 20 µm 0.75 µl 0.6 µm Reverse primer 20 µm 0.75 µl 0.6 µm dntps 2.5 mm 1.25 µl mm Buffer 10 x 2.5 µl 1 x Taq 5 U 0.3 µl 0.06 U/µl DNA template ng/ml 1.0 µl ng/ 25µl H 2 O µl 36

50 With melting temperature (Tm) of various SSR primers ranging from 55 0 C to 60 0 C a touch-down PCR program was set up with: -Initial denaturing: 95 0 C 5 min -Denaturing 94 0 C 30 seconds -Annealing: 60 0 C 1 min ( C drop, 60% RAMP) 10 cycles 55 0 C -Extension: 72 0 C 1 min -Denaturing 94 0 C 30 seconds -Annealing: 55 0 C 1 min -Extension: 72 0 C 1 min 25 cycles -Final extension: 72 0 C 30 min -Holding: 15 0 C With the primer pair CFGA 547a, there was a significant difference of melting temperature (provided by supplier) between forward and reverse primer (approximately 11 0 C). Different PCRs and programs were set up for this primer pair but no amplification was obtained AFLP markers AFLP adapters and primers All sequences of adapters and primers followed Vos et al. (1995). AFLP adapters consist of a core sequence and an enzyme-specific sequence. The structure of the forward and reverse EcoRI-adapter is: Forward: 5-CTCGTAGACTGCGTACC CATCTGACGCATGGTTAA-5 Reverse The structure of the forward and reverse MseI-adapter is: 37

51 Forward: 5-GACGATGAGTCCTGAG TACTCAGGACTCAT-5 Reverse AFLP primers consist of three parts: a core sequence, an enzyme specific sequence (ENZ) and a selective extension (EXT). This is illustrated below for EcoRI- and MseI-primers with three selective nucleotides (selective nucleotides shown as NNN): CORE ENZ EXT EcoRI 5-GACTGCGTACC AATTC NNN-3 MseI 5-GATGAGTCCTGAG TAA NNN-3 Adapters and primers were synthesized by Proligo Primers and Probes (Proligo Australia Pty Ltd) except some labelled EcoRI with 2-3 additional bases which were available at hand. In this study, all EcoRI were labelled with fluorescence called NED or 6-FAM (in the G5 dye set used for ABI 3730 including 6-FAM, VIC, NED, PET and LIZ-standard internal ladder). In accordance with the standard protocol, MseI primers were nonfluorescent DNA digestion and ligation One hundred and fifty to one hundred and eighty ng of genomic DNA were digested with 0.4U/µl of each restriction enzyme EcoRI and MseI (New England Biolabs Inc, 20,000 and 50,000 U/ml, respectively), followed by the ligation of the MseI and EcoRI doublestranded adaptors in the presence of 0.02U/µl T4 DNA ligase (Fermentas Life Sciences, 1,000 U). To check the efficiency of cutting of each enzyme, EcoRI and MseI were used to digest the genomic DNA separately then run on agarose 1% with the size standard Marker II (Roche Applied Science). The quality and quantity of DNA is a critical factor for the AFLP technique. MseI enzyme cut well as it is a frequent cutter. However, as EcoI enzyme is a rare cutter, the efficiency of cutting was not as high as MseI. A mixture to a final concentration of 0.4 U/µl EcoRI (or MseI), 1 x T4 ligase buffer, 1 x BSA, NaCl 0.05M and 3 µl of DNA in the total of 30 µl was made and incubated in the water bath at 37 0 C overnight. When the digestion reaction went well, the digestion and ligation reaction could be combined for convenience (Table 2.7) 38

52 Table 2.7 Digestion and ligation reaction in AFLP technique Reagents Stock 30 µl mix (µl) Final concentration EcoRI 20U/µl U/µl MseI 50U/µl U/µl T4 ligase 1 U/µl U/µl T4 ligase buffer 10 x x BSA 100 x x NaCl 0.5 M M Eco RI adaptor (Forward) 10 mm µm Eco RI adaptor (Reverse) 10 mm µm Mse I adaptor (Forward) 10 mm µm Mse I adaptor (Reverse) 10 mm µm DNA ng/µl ng/µl H 2 O AFLP pre-selective and selective PCRs The products of digestion and ligation were diluted in the ratio of 1:3 with autoclaved water. A pre-selective PCR was run for the annealing of one additional nucleotide to the adapters. Reagents and final concentration used in the pre-selective reaction were: 0.28 µm of each EcoRI primer with additional base A and MseI primer with additional C, Platinum Taq, MgCl 2 and buffer: 0.03 U/µl, 1.5 mm and 1 x, respectively, dntps 0.2 mm and 5 µl of DNA ligase, autoclaved water was added to make up to the total of 25 µl PCR reaction. PCR was programmed: 2 min. at 72 0 C for the initial denaturing followed by 20 cycles with the following cycle profile: a 20 sec. DNA denaturation step at 94 C, a 30 sec. annealing step at 56 0 C, and a 2 min. extension step at 72 0 C. The last extension step was 60 0 C for 30 min. and pre-selective PCR product was held at 15 0 C. The pre-selective PCR product was checked on agarose gel for amplification and then diluted 20 fold for the selective PCR with 2-3 additional nucleotides. The selective PCR reaction used the same amount of reagents above except µm of each EcoRI primer with additional 2-3 nucleotides and MseI primer with 3 additional nucleotide bases, and autoclaved water was 39

53 added to make up to the total 25 µl PCR and ran using the cycling conditions set out in table 2.8. Table 2.8 PCR program for selective amplification step in AFLP technique. 2 cycles cycles cycles 2 holds 94 0 C 65 0 C 72 0 C 94 0 C 64 0 C 72 0 C 94 0 C 44 or 46 0 C 72 0 C 60 0 C 15 0 C 20 s 30 2 min 20 s 30 2 min 20 s 30 2 min 30 min Drop 1 0 C each cycle Ta depends on Mse primer The reproducibility of AFLPs was examined by varying the amount of pre-selective PCR product and the amount of EcoRI +3 (three additional nucleotides) and MseI +3 primers used up to 5 fold. The difference in these two factors within a certain range (two fold higher) did not affect the reliability of AFLPs, except that the height of the peaks tending to be greater with the higher amount of primers. However, when the differences between the amounts of pre-selective PCR product used in the selective PCR were larger, for instance between 20 fold dilution and no dilution, the results were significantly different (data not shown). The quality of DNA was also crucial for the AFLP technique due to its influence on the efficiency of restriction enzymes. Therefore, both the quality and quantity of DNA were critical factors for the AFLP technique Detection method RAPD and ISSR markers RAPD and ISSR fragments were separated according to their size in a 1.5% agarose (Omnigel-low, Edwards Instrument Co. Agarose (low EEC) EED Sulphate < 0.14%) gel run in 0.5 TBE buffer. A 100 bp ladder (DNA molecular weight marker XIV- Roche Diagnostics GmbH) was also loaded on the gel as size standard. The gel was run on different time and voltage settings depending on the size of electrophoresis apparatus used. The gel was stained with Ethidium Bromide and visualized under UV light SSR markers Microsatellite PCR products were first run on agarose to check the amplification of the PCR, using DNA MW Marker XIII (50-750bp) (Cat# , Roche Applied Science, Germany) as a size standard. For primer pair CM16, PCR products from the NNSW 40

54 samples were separated by capillary electrophoresis on an ABI Prism 310 DNA Analyser and analysed by Genotyper 3.7 NT software. The Vietnam samples were separated by an Applied Biosystems 3730 DNA Analyser and analysed by GeneMapper 3.7 software. Ten samples which were run on the ABI Prism 310 Genetic Analyser were re-run on the ABI 3730 DNA Analyser to check the comparability of the two platforms. The results were identical for every single run. For the other five primer pairs, all were run on ABI 3730 DNA Analyser. The ABI Prism 310 Genetic Analyser and ABI 3730 DNA Analyser are laser-induced fluorescence capillary electrophoresis systems. Automated fluorescent scanning detection of DNA enabled separation and quantification of DNA fragments (Fig. 2.4). Fluorescent dyes attached to the fragments via labelled primers emit light at specific wave lengths which can be separated by a spectrograph according to emission frequency. ABI 3730 DNA Analyser uses a 48 capillary array and 36 cm capillary length, and the samples were run for two hours at 8,000 volts. Fig. 2.4 SSR alleles on one locus amplified by CM16 primer pair visualised on the ABI 3730 capillary electrophoresis system AFLP markers Similar to SSR techniques, AFLP PCR products were first run on agarose to check the amplification by PCR and then run on the ABI Following capillary electrophoresis, Genotyper and GeneMapper software (PE Applied Biosystems) were used to aid the 41

55 scoring of the AFLP profiles. The AFLP technique generated a large number of peaks (Fig. 2.5). The multilocus AFLP profiles were scored for the presence (1) or absence (0) of a peak. The height of peaks was not taken into account for reproducibility and for general scoring (McGregor et al. 2000). AFLP data consisted of the presence or absence of peaks on an electropherogram. GeneMapper has the capacity to determine which peaks are common and which peaks are different. A matrix that combines all samples and all unique peaks from the sample set being compared was generated. Each row of the matrix corresponded to one sample and contained ones and zeroes showing the presence or absence of a given peak for that sample, within the size range of 60 bp bp to ensure reproducibility. Fig. 2.5 AFLP fragments amplified by the combination of EcoRI-AAC-NED and MseI-CAA on the ABI 3730 capillary electrophoresis system. 42

56 2.5. Data scoring and analysis SSR markers Following capillary electrophoresis, Genotyper and GeneMapper software (PE Applied Biosystems) were used to aid the scoring of the SSR data. SSR results were scored based on presence and absence of peaks representing for alleles at each locus. - Molecular data format: For the SSR loci in diploid plants, homozygous and heterozygous genotypes were inferred based on the number and size of alleles per locus (one allele in homozygote and two alleles in heterozygous). However, as mentioned above (refer to section 2.5.1), in the case of tetraploids, like C. arabica the number of alleles could be potentially as many as four alleles per locus. The highest number of alleles per locus per single individual in this study was three (at loci CM2 and CM16) (refer to Table 3.4). However, C. arabica is an allotetraploid species exhibiting disomic inheritance (Orozco-Castillo et al. 1994) or diploid-like meiotic behaviour (Lashermes et al. 2000a; Lashermes et al. 2000b) which can be treated correctly only if alleles from different homologous genomes can be distinguished so that genotypes are treated as diploid data (Hardy and Vekemans 2005). In this study, because some loci retained three or two alleles while the ancestral parents genome is uncertain, the determination of heterozygosity and homozygosity became challenging. In addition, available software only supports diploid species analysis. Therefore, microsatellite data was formatted as dominant data in which each allele was treated as a locus and scored as present (1) and absent (0) for only polymorphic marker loci (Bohn et al. 1999; Manifesto et al. 2001; Medini et al. 2005; Montemurro et al. 2005). This data scoring method for SSR might result in a loss of information when genotypes are highly heterozygous (Maguire et al. 2002), however, with predominantly inbred lines like C. arabica, this problem is likely to be minimised (Powell et al. 1996a; Pejic et al. 1998). - Data analysis: To assess the genetic variability and relationship among individuals within and between varieties, SSR data was scored and analysed using the following methods: + Genotypes of all accessions studied were scored based on the allelic pattern of each individual at each locus represented by number and the size of fragments in order to determine the relationship between individuals within varieties and between varieties. + Measurement of genetic variability: Parameters such as band frequency, private bands and mean heterozygosity, molecular variance within and between populations were 43

57 calculated by GenAlEx software (Peakall and Smouse 2005b) to measure the genetic variation within and among cultivars. + Cluster analyses: To investigate the relationship among individuals within and between cultivars, dendrogram and principle coordinate analysis were employed using NTSYS and GenAlEx softwares. The dendrogram was generated based on Nei 72 genetic distance matrix (Nei 1972) and Neighbour Joining method in NTSYSpc 2.10x software (Rohlf 2002) AFLP markers Following capillary electrophoresis, GeneMapper software (PE Applied Biosystems) was used to aid the scoring of the AFLP data. Because AFLPs are dominant markers, their DNA fragments were scored based on the presence (1) or absence (0) of the peak at only polymorphic loci (Remington et al. 1999; Lombard et al. 2000) to create binary matrices Combined SSR and AFLP data In order to obtain a finer genetic structure of the population and a better comparison between the two markers types, the binary data sets of SSR and AFLP were combined into a single data file and analysed as follows. For combined data, the codominant SSR data was simply coded as binary data at only polymorphic alleles so as to match the binary dominant data set of AFLP (Bohn et al. 1999; Manifesto et al. 2001; Medini et al. 2005). Each polymorphic allele of SSR data was treated as a locus (Bohn et al. 1999; Manifesto et al. 2001; Medini et al. 2005). As will be seen in the data analysis following, the SSR analysis clearly describes the fundamental genetic partitioning between coffee varieties and begins to reveal genetic relationships regarding some of the phenotypic off-types. With the combination of SSR and AFLP data the finer genetic structure of the population under study was revealed. This confirmed, as expected in the project objectives (section 1.6.2), that a level of genotypic variation exists within single varieties of NNSW coffee plantations). Almost all previous studies on C. arabica were based on the Unweighted Pair Group Method using Arithmetic Averages (UPGMA) when generating dendrograms. However, Chaparro et al. (2004) avoided UPGMA in making their phylogenetic tree. The argument against UPGMA is that UPGMA assumes a constant rate of evolution (Weir 1990; Hedrick 44

58 2000). This is not always the case in every organismal system, hence UPGMA is criticised for lack of consistency (Bruno et al. 2000), although it is the simplest method for constructing trees, sometimes called phenograms, and it was originally used to represent the extent of phenotypic similarity for a group of species in numerical taxonomy (Nei 2000). Saitou and Nei (1987) found that UPGMA showed a poor performance when compared with other tree-making methods in terms of proportion of correct tree and distortion index of two model trees. In contrast, the Neighbour Joining (NJ) method showed high performance in obtaining the correct unrooted and rooted trees in comparison with those obtained from other tree-making methods and is applicable to most types of evolutionary distance data (Saitou and Nei 1987; Saitou and Imanishi 1989). When exact genetic distance matrices are given, NJ can reproduce the correct tree with elegance and speed (Bruno et al. 2000). Therefore, NJ analysis was performed in the present study. To run NJ analysis, Nei s distance matrix (Nei 1972) was calculated based on the principle of pairwise distance. Whilst Nei (1972) suggested a dissimilarity coefficient based on mutation and genetic drift, often known as Nei s standard distance or Nei 72, it should be noted that this genetic coefficient was developed based on the infinite allele model (Kimuraz and Crow 1964) assuming that an ancestral population split into various subpopulations, which diverged due to genetic drift and mutation. If the mutation-drift balance is maintained throughout the evolutionary process, it will be assumed that selection is absent and the dissimilarity is not large (Reif et al. 2005). If one can assume the infinite allele model, then standard genetic distance of Nei (1972) is suitable for investigating phylogenetic relationships among populations (Reif et al. 2005). Although such an assumption would seem at odds in a situation of a plantation population of a cultivated species, albeit propagated from seed, the most commonly used genetic distance measure like Nei 72, with its robustness, could yield a dendrograms which faithfully represents the relative relationships between groups of individuals within the population (Hedrick 2000; Kalinowski 2002). NJ, like UPGMA, consists of two main steps that are repeated until a tree is complete. The first step consists of choosing a pair of taxa to be joined, i.e., replaced by a single new node 45

59 representing their immediate common ancestor. In the second step, distances from the new node to all other nodes are inferred (Bruno et al. 2000). It was analysed as follows: - To test the correlation between SSR and AFLP data, the Mantel test was applied using GenAlEx software. - To measure the informativeness of each type of marker, polymorphic loci, number of bands, number of private bands of each type of the markers and comparison between two marker types were calculated by GenAlEx software. - To measure genetic variation within and among cultivars, heterozygosity and genetic dissimilarity, principle coordinate analysis and dendrogram were employed using GenAlEx and NTSYS softwares. 46

60 CHAPTER III RESULTS 3.1. RAPD results A total of 20 primers were applied in this study (Table 3.1) and generated 65 RAPD primer/restriction enzyme (RAPD/RE) combinations. Ten out of twenty primers were run with genomic DNA digested by five different restriction enzymes (RE), of which 5 primers were also run with uncut genomic DNA, to compare the banding patterns amounting to 55 different RAPD/RE combinations. The other ten primers were only run with prior digestion of genomic DNA by HindIII. The number of fragments generated by any one primer varied from as few as one (OPA07, uncut DNA) to as many as thirteen bands (OPK14 uncut, OPK14 with HindIII and OPK14 with EcoRI). The size of fragments ranged from 300 bp to 2500 bp. For some primers, there was no difference between nondigested and digested genomic DNA before amplification, such as primer OPC10 with and without digestion with HindIII before PCR (Table 3.1). Fifteen out of 65 RAPD/RE combinations were clear enough for scoring. The remaining fifty RAPD/RE combinations produced 333 bands of which 10 bands (3%) generated from just nine RAPD/RE combinations (18%) were polymorphic. However, five out of these nine combinations were generated from the same primer (OPN20) digested with different restriction enzymes. Whilst these combinations showed polymorphic bands (present or absent) amongst samples, the different template restriction events all produced fragments of similar size when present. Four remaining polymorphic combinations came from two other primers (OPC07 and OPA07) in which two combinations generated from primer OPC07 (OPC07-HindIII and OPC07-EcoRI) shared the same polymorphic locus. Due to its reliability and confidence in scoring, only one primer and enzyme combination (OPN20-HindIII) was tested again for all 84 samples in NNSW and 11 samples from Vietnam to assess the polymorphism in the whole sample population. The number of individual bands exhibiting polymorphism in this study (3%) is much lower than that among 14 elite C. arabica (68%) reported by Sera et al. (2003). Likewise, the number of polymorphic markers in this study (18%) was lower than that among coffee progenies and cultivars (67.67%) reported by Silveira et al. (2003) when applying the same technique. 47

61 For some primers, the RAPD profiles obtained in C. arabica were not affected by restriction digestion of the template DNA (eg. OPN20, OPC07). This is also observed in wheat (Koebner 1995) and in other C. arabica cultivars (Sera et al. 2003). Other primers, however, produced a fewer number of fragments in uncut than cut template (UBC220, OPA07). Conversely, some primers like OPK14 produced fewer bands when the genomic DNA was digested. There was no clear difference in the number of bands on the gel between rare cutters and frequent cutters. Primer OPN20 associated with HindIII, EcoRI and BamHI restriction had the same banding patterns as with uncut template (Fig. 3.1), while primer UBC220 exhibited a difference between uncut and cut with HindIII, which gave a similar profile to EcoRI and BamHI. Therefore banding patterns of the same primer with different RE can be similar or different depend on the combination of restriction site and annealing site. A claimed shortcoming of the RAPD technique is that amplification results may vary between PCR reactions. Whilst a certain number of bands remain unchanged between each PCR reaction, others may be well amplified in one PCR but weakly amplified in another, making the banding pattern inconsistent and difficulty to score consistently. A proposed explanation for this inconsistency of amplification is that RAPDs utilising relatively low annealing temperatures coupled with variable DNA quality may exacerbate reproducibility problems (Micheli et al. 1994). Furthermore, the present study applied a modified RAPD technique in which genomic DNA was cut by restriction enzymes. DNA purity therefore becomes more crucial since it affects the cutting efficiency of the enzyme. 48

62 Fig. 3.1 The banding pattern of primer OPN20 using genomic template DNA digested with different restriction enzymes (each enzyme digested four different DNA samples) 1150 bp Fig. 3.2 The banding pattern of RAPD primer OPN20 using HindIII digested genomic template DNA showing polymorphic locus at 1150 bp. 49

63 The nine polymorphic markers differentiated seven varieties into two groups, SL14, SL34 and K7 in one group and the remainders belong to the other. No inter- or intra-variety variation was detected (Fig. 3.2). Neither could any association between marker polymorphism and morphological traits be determined. Although marker OPN20-HindIII was screened for the whole sample population, only one polymorphic locus was detected with the allele absent in CRB, Catuai, Arusha and Mundo Novo but present in K7, SL14 and SL34 (Fig. 3.2). Table 3.1 RAPD amplification results No Primer Enzyme Number of Number of Polymorphic loci Amplified polymorphic (bp) fragments fragments 1 OPN20 uncut HindIII EcoRI RsaI BamHI HaeIII UBC 220 uncut 2 0 HindIII Not reproducible EcoRI Not reproducible RsaI 4 0 BamHI 7 0 HaeIII OPC07 uncut 7 0 HindIII , 820 EcoRI , 820 RsaI 3 0 BamHI 9 0 HaeIII OPK14 uncut 13 0 HindIII 13 0 EcoRI 13 0 RsaI 8 0 BamHI 12 0 HaeIII OPA07 uncut 1 0 HindIII 6 0 EcoRI unscorable RsaI 8 0 BamHI HaeIII cont d/. 50

64 No Primer Enzyme Number of Number of Polymorphic loci Amplified polymorphic (bp) fragments fragments 6 UBC16 HindIII unscorable EcoRI unscorable RsaI unscorable BamHI unscorable HaeIII OPP18 HindIII 9 0 EcoRI unscorable RsaI unscorable BamHI unscorable HaeIII unscorable 8 OPI20 HindIII 9 0 EcoRI 9 0 RsaI unscorable BamHI 9 0 HaeIII unscorable 9 OPN18 HindIII 4 0 Hind, Eco Rsa have EcoRI 4 0 same banding patterns RsaI 3 0 BamHI unscorable HaeIII Z14 HindIII unscorable EcoRI 4 0 RsaI unscorable BamHI unscorable HaeIII OPB02 HindIII OPC10 HindIII OPC15 HindIII OPF05 HindIII OPK18 HindIII OPS15 HindIII OPX09 HindIII OPX20 HindIII UBC212 HindIII UBC217 HindIII 3 0 Due to the inconsistency and low polymorphism of RAPDs, they were not included in further analysis, so other marker systems were sought ISSR results Ten single primers and twelve paired-primer combinations (see Table 3.2) detected polymorphism in 11 the individuals representative of seven varieties as described in 51

65 method section. Two of twenty-two markers screened were unscorable (R03 and R03+R04). Three single primers (R06, R07 and R08) failed to amplify a PCR product. Only one out of 17 (5.9%) primers (R02) showed polymorphism. The seventeen single and combined primers produced 113 bands in which only 1 was polymorphic (0.9%) in the 84 NNSW and 11 Vietnamese samples were analysed. It is not surprising that the results here suggest extremely low levels of polymorphism compared with those observed in eight species of Coffea and six interspecific hybrids (96.5%) (Ruas et al. 2003) because the coffee samples in this study belong to cultivars of C. arabica species while those of Ruas et al. (2003) came from different species of Coffea. The assessment of genetic integrity of the nuclear, mitochondrial and chloroplast genomes among micro-propagated plants and a single mother plant also showed relatively low level of polymorphism, 17.6% for polymorphic primers and 6.7% for polymorphic ISSR loci (Rani et al. (2000), which is therefore consistent with the current study data. ISSRs performed similarly to RAPDs in the number of amplified fragments, with as few as 1 band (R09) to as many as 11 bands (R01 + R02, R01 + R05) and the size of fragments ranging from 100 bp to 2642 bp. R02 detected 6/7 polymorphic loci (polymorphic loci/total loci) in the samples surveyed by Ruas et al. (2003) but in our study it revealed only 1/7 polymorphic loci and was the only primer to show polymorphism. A number of primers that showed relatively high polymorphism (R01+R03-20/20, R04+R05-16/16, R01+R02-15/15 and R05-10/10) in the samples surveyed by Ruas et al. (2003) showed no polymorphism in this study. The presence of a 480 fragment in primer R02 enabled the C. arabica cultivar CRB accessions to be distinguished from K7 accessions examined in this study (Fig. 3.3). Like RAPDs, ISSR markers were not informative enough to detect the differences between varieties and also not reproducible between PCRs. ISSR markers therefore did not employed further in this analysis and other more informative marker systems were sought. 52

66 480 bp Fig. 3.3 Polymorphism revealed by ISSR (R02) markers distinguishes between K7 and CRB at the fragment length of 480 bp. Table 3.2 ISSR primers and amplifications No Primer Sequence Number of Number of Fragment Polymorphic 5-3 amplified polymorphic size range Locus fragments fragments (bp) (bp) 1 R01 (GA) 9 -T R02 (GA) 9 -C R03 (GGAT) 4 unscorable 4 R04 (GACA) R05 (GATA) R06 (GA)9 NA* 7 R07 (TC)9 NA* 8 R08 (TAG)6 NA* (original paper used (TAG)4) 9 R09 (CCTA) R10 (GGTA) R01+R R01+R R01+R R01+R cont d/.. 53

67 No Primer Sequence Number of Number of Fragment Polymorphic 5-3 amplified polymorphic size range Locus fragments fragments (bp) (bp) 15 R02+R R02+R R02+R R03+R04 unscorable 19 R03+R R04+R R02+R R02+R *NA= no amplification 3.3. SSR markers One of six primers did not amplified although different PCR reactions were applied. Four of five SSR primers revealed polymorphism in the 84 NNSW samples and 11 Vietnam samples. The repeat motifs were dinucleotides. The PCR products were applied multiplex loading for the reason of cost. The number of alleles per locus ranged from one (CFGA189 NED) to six (CM2). The fragment size range was from 96 bp to 278 bp. CM2 was the most informative marker that gave 6/6 polymorphic alleles per locus (Table 3.3). The highest number of alleles per locus per single individual was three (CM2 and CM16). Table 3.3: Primer information and the results of SSR amplification Primer Repeat motif Product size range (bp) Na Npa CM2 FAM (AC) 10 (AT) CM8 TET (GA) 7 (GT) CM16 HEX (GA) CFGA189 NED (AG) CFGA502 FAM (AG) CFGA547a VIC (AG) 18 No amplification Na= Number of alleles Npa= Number of polymorphic alleles per locus. 54

68 Table 3.4: Genotypes of all accessions studied at four loci of microsatellites. LOCUS1 LOCUS2 LOCUS3 LOCUS4 LOCUS1 LOCUS2 LOCUS3 LOCUS4 Accessions Genotype Genotype Genotype Genotype Accessions Genotype Genotype Genotype Genotype K7-CA01* A C A A CRB-C1 B A A B K7-CA02* A C A A CRB-C2 B A A B K7-CA03 A C A A CRB-K1 B A A B K7-CA04* A C A A CRB-K2 B A A B K7-CA05 A C A A CRB-K3 B G A B K7-CA06* A C A A CRB-K4 B A A B K7-CA07 A C A A CRB-K5* A A A A K7-CA08 A C A A CRB-K6 B A A B K7-CA09 A C A C CRB-K7 B A A B K7-CA10* A C A A CRB-K8 B A A B K7-CA11* A C A A CRB-K9 B A A B K7-CA12 A F A B CRB-K10 B A A B K7-CA13 A C A C CRB-K11 B A A B K7-CA14 A E A A CRB-K12 B A A B K7-CB01 A C A A CRB-K13* A A A A K7-CB02* A C A A CRB-K14 B A A B K7-CB03 A F A A CRB-K15 B H A B K7-CB04 A C A A CRB-K16 B A A B K7-CB05 A C A A CRB-K17 B H A B K7-CB06 A C A A CRB-K18 B A A B K7-CB07 A C A A CRB-K19 B A A B K7-CB08 A C A A CRB-K20* A A A A K7-CB09 A C A A BourbonVN1 B A A A K7-CB10 A C A A BourbonVN2 A A A B K7-K01 A C A A BourbonVN3 A B A B K7-K02 A C A A BourbonVN4 B A B A K7-K03 A C A A Arusha1 B A A B K7-K04 A C A A Arusha2 B A A B K7-K05 A C A A Catuai1 B A A A K7-K06 A C A A Catuai2 A A A A K7-K07 A C A A CatuaiVN1 B A A B K7-K08 A C A A CatuaiVN2 B A A B K7-K09 A C A A CatuaiVN3 B A A B K7-K10 A C A A CatuaiVN4 B A A B K7-K11 A C A A SL14 A C A A K7-K12 A C A A SL14VN A A A B K7-K13 A C A A SL34 A C A A K7-K14* A C A A MNovo1 B A A A K7-K15 A C A A MNovo2 B A A A K7-K16 A C A A MNovoVN A H A A Table 3.4 contd/.. 55

69 Accessions LOCUS1 Genotype LOCUS2 Genotype LOCUS3 Genotype LOCUS4 Genotype K7-K18 A C A A K7-K19 A C A A K7-K20 A C A A K7-M01 A C A A K7-M02 A C A A K7-M03 A C A A K7-M04 A C A A K7-M05 A C A A K7-M06 A F A B K7-M07 A C A A K7-M08 A I A B K7-M09 A C A A K7-M10 A C A A K7-VN B F A A Locus 1: Genotype A: allele 96 bp, 106 bp** Genotype B: allele 96 bp, 106 bp, 110 bp Locus 2: Genotype A: allele 195 bp, 217 bp Genotype C: allele 197 bp, 219 bp Genotype E: allele 197 bp, 219 bp, 221 bp Genotype F: allele 197 bp, 217 bp Genotype G: allele 195 bp, 215 bp, 217 bp Genotype H: allele 195 bp, 217 bp, 219 bp Genotype I: allele 197 bp, 217 bp, 219 bp Locus 3: Genotype A: allele 236 bp, 240 bp Genotype B: allele 234 bp, 240 bp Locus 4: Genotype A: allele 181 bp Genotype B: allele 182 bp Genotype C: allele 181 bp, 182 bp * accessions expressing morphological difference compared with other individuals in the same variety. ** fragment size of amplicon or allele size (bp) Letters in red and underline indicate the distinct genotypes Letters in black indicate the common group - representing genotypes. Marker genotypes of all studied accessions were scored at four different loci corresponding to four primer pairs were scored, namely Locus1 = CM16, Locus 2 = CM2, Locus 3 = CFGA 502 and Locus 4 = CM8. Marker genotypes were defined based on the number and the length of the alleles generated by each SSR marker, which is described in table 3.4. At locus 1 (primer pair CM16), genotype A (alleles 96 and 106 bp) was typical for K7, SL14 and SL34, while genotype B (alleles 96, 106 and 110 bp) was the predominant genotype for CRB, Arusha, Catuai and Mundo Novo (Table 3.4 and Fig. 3.4). However, a single K7 individual collected from Vietnam possessed the B genotype. Similarly, one Catuai and three CRB accessions from NNSW and one Mundo Novo and two Bourbon individuals from Vietnam exhibited the A genotype. 56

70 Locus 2 (primer pair CM2) was most informative with 7 marker genotypes for the accessions studied. C was the typical genotype for K7, SL14 and SL34, except SL14 from Vietnam, while A was the predominant genotype for CRB, Arusha, Catuai and Mundo Novo. Among K7, there are six individuals with distinct genotypes named E, F and I, but have a typical K7 phenotype. Five individuals in variety CRB also possessed discrete genotypes called B, G and H. These plants were indistinguishable from others of their variety. Locus 3 (primer pair CFGA 502) was least informative with genotype A in common for all accessions studied except BourbonVN 04 from Vietnam which had B genotype. At locus 4 (primer pair CM8), allele 181bp was scored as A for K7, SL14 (except SL14-VN), SL34, Catuai from NNSW (2 individuals) and all Mundo Novo individuals. Allele 182bp was scored as B for CRB, Arusha, and 4 Catuai individuals from Vietnam. Two K7 accessions appeared to have 2 alleles at this locus scored as C. Some K7 individuals exhibited locus 4 s genotype from CRB, and vice versa. 57

71 Fig 3.4 Polymorphisms among varieties revealed by SSR markers (CM16). Num ber of bands K7 CRB Catuai Heterozygosity No. Bands No. Bands Freq. >= 5% No. Private Bands Mean Heterozygosity Cultivars Fig. 3.5 Band frequency, private bands, mean heterozygosity across K7, CRB and Catuai. 58

72 The number of bands (since SSR data were formatted as dominant data) in K7 and CRB was similar (Fig. 3.5). However, the number of private bands in K7 was twice as many as CRB (2 and 1, respectively). This may be a result of the sample size of K7 being approximately 2.5 times bigger than CRB (54 and 22, respectively), so the chance of detecting rare alleles in K7 is higher. The number of less common bands ( 25% and 50%) was zero, suggesting that common alleles were shared among individuals within variety. Heterozygosity is an important measure of genetic variability of a population regardless of whether the species is a self-fertilizer or outbreeder or whether it is diploid or polyploid (Nei 1975; Nei et al. 1976; Weir 1990; Hedrick 2000). The mean heterozygosity was highest in Catuai, followed by CRB and K7 (0.116, and 0.084) indicating that CRB is more diverse than K7 but this difference appeared insignificant based on the overlapping error bars (Fig. 3.5). Pairwise genetic identity (I=1-genetic diversity, Nei (1987)) between K7 (n=54) and CRB (n=22) was 0.46, suggesting a moderate level of genetic diversity between the two cultivars. Analysis of Molecular Variance (AMOVA) for the SSR data of 76 genotypes from K7 (54) and CRB (22) showed that 89% and 11% of the genetic variation was found between and within population, respectively (Table 3.5 and Figure 3.6.). These data were supported by a ΦPT = 0.89 which measures the magnitude of genetic differentiation among populations. ΦPT is calculated as the proportion of the variance among populations, relative to the total variance (Peakall and Smouse 2005a). It is analogous to Fst, but Fst is calculated in the case of codominant data while ΦPT is calculated for binary data (Peakall and Smouse 2005a). Within population 11% Between populations 89% Fig. 3.6 Molecular Variance based on the calculation of PhiPT (ΦPT ) for the SSR data of 76 genotypes from K7 (54) and CRB(22). 59

73 Table 3.5: Results of Analysis of Molecular Variance (AMOVA). Source of variation df SS % of variation Between population *** Within population ΦPT = 0.89, ***P < df: Degrees of freedom SS: Sum of square NS: Non-significant at Prob : Level of significance determined by a 999 permutation test. The AMOVA suggested that the variation was very large between K7 and CRB but smaller within each population. Phylogenetic analysis using NJ based on the four microsatellites showed a clear separation between the two sub-species or two genetic groups including the Bourbon-derived genotypes and the Typica-derived genotypes (Fig. 3.7). However, there were few phylogenetically informative characters to make intra-variety -differentiation. A dissimilarity matrix of Nei 72 genetic distance values was used for cluster analysis using NJ method. The resulting dendrogram is presented in Fig NJ analysis grouped the 95 accessions studied into two main clusters that generally agree with their origins and pedigrees with cluster I having 66 genotypes and cluster II having only 29 genotypes. Cluster I was divided into two subclusters: subcluster Ia containing 64 genotypes (almost all K7, SL14, SL34 from NNSW and three Mundo Novo (two from NNSW, one from Vietnam), three CRB and two Catuai from NNSW) and subcluster Ib containing only two Bourbon genotype from Vietnam). The second cluster consisted of 6 varieties and was highly informative with separated branching in the dendrogram. There is one subcluster comprising all individuals from CRBs (except the special three in the first cluster - see below), 2 from Arusha, 4 from Catuai and a few of others (1 K7, SL14 VN and Bourbon VN02) and one outlier - K7. The failure of this genotype to cluster is most likely the result of the limited SSR data set. In summary, there were two polytomies (unresolved branches) in the dendrogram. The first polytomy was lined up with 49 genotypes in which 47 genotypes belonged to K7 variety and the other two were SL14 and SL34 from NNSW. 60

74 The second polytomy was lined up with 22 genotypes in which 16 genotypes belonged to CRB variety and the other six were two Arusha from NNSW and four Catuai from Vietnam. It was observed that only two K7 individuals belonged to cluster II while the remaining of K7 were gathered into the first cluster. These two K7 have no morphological difference comparing to other K7 accessions. Conversely, three CRB individuals belonged to cluster I while the remaining were grouped into cluster II. Unlike the two K7 mentioned above, these three CRB also have morphological dissimilarity to other CRBs. Dissimilarity coefficients between all possible pairs of genotypes ranged from 0 to 0.53 for the whole set of samples studied. When generated separately for each population, the coefficients ranged from 0 to 0.31 among individuals within CRB variety and from 0 to 0.23 among individuals within K7 variety (dendrogram not shown relationship between K7 & CRB individuals were as revealed in dendrogram in Fig. 3.7). These low values reflect the high proportion of common alleles among genotypes within each group and are consistent with the data presented above. 61

75 Fig. 3.7 Phylogenetic tree constructed by dissimilarity matrix (Nei 72) and Neighbour Joining method (NTSYSpc 2.10) using SSR data. 62

76 This low differentiation might be due to the limited number of SSR markers used in this chapter, which would be improved by combining with the AFLP data that will be described in section AFLP markers A total of 22 primer combinations were run on the ABI 3730 with 14 individuals represented for 8 varieties. K7: 3 individuals Arusha: 1 individual CRB: 2 individuals SL14: 2 individuals Bourbon: 1 individual SL34: 1 individual Catuai: 2 individuals Mundo Novo: 2 individuals Five out of twenty-two primer combinations showed polymorphism, of which only the two most informative combinations were screened across all individuals. Reproducibility tests for these chosen combinations were performed for four samples using five-fold and 10-fold dilutions of pre-selective PCR product and µm and 0.05 µm of each Eco +3 and Mse +3 primer. The other three combinations were excluded from further analysis as only one reliable polymorphic locus for each of these combinations was found. The AFLP results are presented in Table 3.6. The number of fragments amplified ranged from one (EcoRI-ACA and MseI-CTG combination) to 78 (EcoRI-AAC and MseI-CAA combination). Scoring the fragments was restricted to those fragments of less than 400 bp. The size range was from around 60 bp up to bp for samples with good amplification and below 300 bp for poor and moderate amplification. 63

77 Table 3.6 AFLP combinations and the number of fragments amplified Labelled EcoRI +2-3 MseI +3 No of bands NPB Size range (bp) Amplification capacity AC- FAM CTA Moderate amplification AT- NED CTA Moderate amplification AAC- NED CAA Good amplification CAC Moderate amplification CAG Bad amplification CAT Good amplification CTA Good amplification CTC Moderate amplification CTG Good amplification CTT Good amplification ACA- FAM CAA , 156 Bad amplification CTA Bad amplification CTG Bad amplification CTT 0 0 No amplification ACT- FAM CTA Poor amplification ACC- NED CAG Moderate amplification CAT Poor amplification CTA Good amplification AGC-NED CAA Moderate amplification CTA Bad amplification CTG 0 0 No amplification CTT 0 0 No amplification NPB = Number of polymorphic bands. Letters in red: The most informative primers and their results 64

78 The combination of EcoRI-AAC with MseI-CTT and EcoRI-AAC with MseI-CAA generated a number of polymorphic loci (Fig. 3.8). The former combination gave 56 loci in total, of which seven were polymorphic loci. The latter combination had fewer informative loci (only three variable loci out of 78 loci detected). Using the EcoRI-AAC and MseI-CAA combination, Steiger et al. (2002) observed 57 polymorphic bands with a sample of 61 accessions from 6 cultivars while in the present study only 3 polymorphic bands were evident. Given the limited number of polymorphic markers and combinations found (10 and 2, respectively), AFLP markers were analysed in combination with SSR data for further analysis as discussed in the following section. 65

79 Fig. 3.8 Polymorphisms among varieties revealed by AFLP markers (EcoRI-AAC and MseI-CAA combination) 66

80 3.5. Combined SSR and AFLP data Correlation between AFLPs and SSRs Before AFLP and SSR data were combined, a Mantel test was applied to assess the correlation between two types of data. Ten polymorphic AFLP loci from two combinations were scored as binary data and subjected to GenAlEx software to obtain the genetic distance matrix. The genetic distance matrices obtained through AFLP and SSR analyses were compared using Mantel test with permutational testing procedures (GenAlEx software). The values of the Mantle test showed a significant and positive correlation (r = 0.24, P <0.001) between individual pairwise genetic distance matrices for both AFLP and SSR data set. This is in agreement with former observations, where similar results were obtained with AFLP and SSR at the same taxonomic levels (Coart et al. 2003; Prakash et al. 2005) or higher taxonomic levels (Maguire et al. 2002). These results indicate a good fit of the data obtained by two marker types and thus the results are comparable across the whole data set. The AFLP data were therefore combined with SSR for further analysis. The approach to presenting SSR data as dominant was explained in section Genetic variation between varieties To measure the genetic variation, the parameters such as percentage of polymorphic loci, total number of bands, number of private bands and mean heterozygosity were calculated based on GenAlEx software. Table 3.7 DNA polymorphism of seven arabica varieties using SSR and AFLP data. Percentage of Polymorphic Loci Data No. Markers K7 CRB Bourbon Arusha Catuai SL M. Novo n=55 n=22 n=4 n=2 n=6 n=3 n=3 SSR 11 64% 45% 73% 0% 27% 55% 18% AFLP 10 80% 80% 40% 10% 20% 30% 10% Combined 21 72% 63% 56% 5% 24% 42% 14% SSR No. of Bands AFLP Combined SSR No. of Private Bands AFLP Combined SSR Mean heterozygosity AFLP Combined

81 The combined marker data included up to 11 SSR and 10 AFLP loci for mainly K7 and CRB varieties. Other additional varieties were also included in the study to test the robustness of markers. The additional varieties showed a different level of polymorphic loci compared to K7 and CRB (Table 3.7) and this may be a result of the small sample size in the additional varieties. However, the presence of polymorphisms in the five additional varieties demonstrated the informativeness of the markers. The percentage of polymorphic loci was lower in SSR (64% for K7 and 45% for CRB) than AFLP data (80% for both K7 and CRB). There was no private AFLP band for any variety and only one private SSR band for K7, CRB and Bourbon. From the combined DNA markers data, genetic dissimilarity was calculated based on pairwise comparison by NTSYS software. In this analysis, only the two main cultivars were included K7 (54 individuals) and CRB (22 individuals). Number of pairs (%) > <0.1 Genetic dissimilarity Fig 3.9 Pair-wise comparison of genetic dissimilarity among individuals between K7 and CRB (76 individuals) using combined data. Fig. 3.9 shows genetic dissimilarity among 76 individuals from K7 and CRB. Around 57% and 24% of the pair-wise comparisons exhibited a genetic dissimilarity less than 0.5 and 0.1, respectively (Fig 3.9). Only 1.3% of the pair-wise comparisons displayed a genetic dissimilarity greater than 1.3. Around 42% of the pair-wise comparisons showed higher genetic dissimilarity values from 0.5 to 1.3. The highest genetic dissimilarity value for all pair-wise comparisons between two varieties was 1.5 (between CRB K11 and K7-CA14). 68

82 The highest genetic dissimilarity value for all pairwise comparisons between two varieties K7 and CRB was 1.5. This value is not particularly high, because the genetic distance calculated in GenAlEx based on Nei (1972) ranges from 0 to. Nei s standard genetic distance (Nei D) is calculated: Nei D = ln (I), where I is Nei s Genetic Identity and its value is ranged from 0 to 1 (Peakall and Smouse 2005a). The pair-wise genetic identity therefore gives a better description of the genetic relationship. The genetic identity described the genetic relationship between K7 and CRB was average (0.641) Genetic variation within varieties The genetic variation at the DNA level among individuals within the K7 cultivar was detected using AFLP markers (Fig & Fig. 3.11). The degree of variation was determined by comparing the genetic dissimilarity in pairs (Fig & Fig. 3.12). There was marked genetic difference between plantations (Fig. 3.11). The mean genetic dissimilarity within K7 was highest in Block C (0.25) followed by Block M (0.20) and Block K (0.09) even though Block M had the smallest sample size. However, the genetic distance values are generally low since the Nei s standard genetic distance value could be from 0 to as explained above. The highest value of genetic dissimilarity for all pair-wise comparisons within Block C (276 pairs), Block M (45 pairs) and Block K (190 pairs) was 0.85, 0.55 and 0.28, respectively. In other words, Block C displayed the highest level of genetic diversity. Block M, despite having the smallest number of samples, still exhibited a higher level of genetic diversity than Block K which was more uniform and had a larger number of samples. 69

83 Fig Polymorphisms within K7 variety revealed by AFLP markers (EcoRI-AAC and MseI-CAA combination) Mean genetic dissimilarity Block C (n=24) Block K (n=20) Block M (n=10) K7 blocks Fig Pair-wise comparison of genetic dissimilarity (Nei 72) within K7 in different blocks using combined data. Error bars represent the standard errors of means. 70